Report IEA-PVPS T13-02:2014

Characterisation of Performance of

Thin-film Photovoltaic Technologies

Outdoor test facility for the comparison of energy yield measurements of different photovoltaic module technologies. Courtesy of TÜV Rheinland, Energie und Umwelt GmbH, Cologne, Germany.

iii

INTERNATIONAL ENERGY AGENCY

PHOTOVOLTAIC POWER SYSTEMS PROGRAMME

Characterisation of Performance of

Thin-film Photovoltaic Technologies

IEA PVPS Task 13, Subtask 3.1

Final Report IEA-PVPS T13-02:2014

May 2014

ISBN 978-3-906042-17-6

Authors:

Timothy J Silverman

National Renewable Energy Laboratory, Golden, CO, USA

Ulrike Jahn

TÜV Rheinland Energie und Umwelt GmbH, Cologne, Germany

Gabi Friesen and Mauro Pravettoni

SUPSI ISAAC, Canobbio, Switzerland

Marco Apolloni

TEL Solar, Trübbach, Switzerland

A Louwen and WGJHM van Sark

Utrecht University, Copernicus Institute of Sustainable

Development, Utrecht, the Netherlands

Markus Schweiger

TÜV Rheinland Energie und Umwelt GmbH, Cologne, Germany

G. Belluardo, J. Wagner, A. Tetzlaff, P. Ingenhoven, D. Moser

EURAC Research Institute for Renewable Energy, Bolzano, Italy

iv

Further contributions and data from:

Anne Gerd Imenes, Teknova, Norway

Lionel Sicot, Guillaume Razongles, CEA INES, France

Tom Betts, CREST, United Kingdom

George E. Georghiou, University of Cyprus

Mike van Iseghem, Didier Binesti, EDF R&D, France

This report is supported by:

Austrian Federal Ministry for Transport, Innovation and Technology (BMVIT) under FFG

contract No. 828105

German Federal Ministry for Economic Affairs and Energy under Contract No.0325194A

(BMWi)

Swiss Federal Office of Energy (SFOE)

“FLASH”, a Perspectief programme funded by Technology Foundation STW (http://stw.nl/en/)

under project number 12172

and

U.S. Department of Energy under Contract No. DE-AC36-08-GO28308 with the National

Renewable Energy Laboratory (NREL)

v

Table of Contents

1 Foreword ........................................................................................................................... 1

2 Executive Summary .......................................................................................................... 2

3 Simulator Performance ...................................................................................................... 3 3.1 Previous work ...................................................................................................... 3

3.1.1 Introduction ........................................................................................................ 3 3.1.2 State of the art in indoor performance characterization of PV modules with solar simulators ...................................................................................................................... 3 3.1.3 Specific issues in thin-film modules .................................................................... 7 3.1.4 References for this section .............................................................................. 10

3.2 Stabilisation issues particular to thin-film PV modules .................................. 11 3.2.1 Introduction ...................................................................................................... 11 3.2.2 Recent results in thin-film stabilisation ............................................................. 11 3.2.3 Improving upon thin-film PV stabilisation methods ........................................... 16

3.3 Performance characterization of thin-film multi-junction PV cells and modules 22

3.3.1 Introduction ...................................................................................................... 22 3.3.2 Theory and definitions...................................................................................... 22 3.3.3 Basics of measurement methods ..................................................................... 24 3.3.4 Acknowledgments ............................................................................................ 29 3.3.5 References for this section .............................................................................. 30

4 Field performance ........................................................................................................... 32 4.1 Introduction ........................................................................................................ 32 4.2 Summary of previous studies in the literature ................................................. 32

4.2.1 Introduction ...................................................................................................... 32 4.2.2 Spectral effects ................................................................................................ 32 4.2.3 Temperature effects ......................................................................................... 34 4.2.4 Transient and seasonal performance changes ................................................ 35 4.2.5 References for this section .............................................................................. 37

4.3 Comparison of Different PV Module Technologies: New Method to Analyse Performance Characteristics Obtained from Field Tests at Different Locations ....... 38

4.3.1 Motivation ........................................................................................................ 38 4.3.2 Approach ......................................................................................................... 38 4.3.3 Results ............................................................................................................ 40 4.3.4 References for this section .............................................................................. 44

4.4 Analysis of spectral effects on outdoor performance ..................................... 44 4.4.1 Spectral response measurements ................................................................... 44 4.4.2 Measuring solar spectral irradiance ................................................................. 45 4.4.3 Analysis of large amounts of spectral data with the APE factor ........................ 47 4.4.4 Correlating SR data and spectral data with module performance ..................... 49 4.4.5 Influence of spectral effects on the energy yield of single-junction modules ..... 50 4.4.6 References for this section .............................................................................. 51

4.5 Modeled spectrum method ................................................................................ 51 4.5.1 Introduction ...................................................................................................... 51 4.5.2 Description of methodology ............................................................................. 51 4.5.3 Results ............................................................................................................ 53 4.5.4 Acknowledgements .......................................................................................... 57 4.5.5 References for this section .............................................................................. 57

vi

List of Figures

Fig. 1: Spread of Pmax at STC (deviation from average of six test laboratories) as measured with different solar simulators and four different crystalline silicon module types ............... 4

Fig. 2: Spread of Pmax at 200 W/m² (deviation from average of six test laboratories) as measured with different solar simulators and four different crystalline silicon module types5

Fig. 3: Spread of Pmax temperature coefficient measured at 800 W/m² (deviation from average of six test laboratories) as measured with different solar simulators and four different crystalline silicon module types. ........................................................................................ 5

Fig. 4: Spread of Pmax at STC (deviation from average of six test laboratories) as measured with different solar simulators and four different thin-film module technologies within the second round-robin ........................................................................................................... 6

Fig. 5: Normalised PV parameters of single junction thin-film devices as measured (i) outdoors, (ii) indoors without spectral mismatch correction, and (iii) indoors with spectral mismatch correction. The variations are relative to the outdoor measurement [21]. .......................... 9

Fig. 6: Normalised power (Pmax) of single-junction thin-film devices (left) and double-junction and hybrid a-Si (right) module, measured as a function of I-V sweep-time. The dashed line indicates a typical sweep-time used for the LAPSS solar simulator [21]. ..................... 9

Fig. 7: Electrical stabilisation of a CIGS specimen during light-soaking at 1000 W/m² and a module temperature of 58.4°C ........................................................................................ 12

Fig. 8: Irreversible degradation of a CIGS specimen caused by light-soaking at 83.7°C module temperature, online-monitored module parameters at 1000 W/m² ................................... 12

Fig. 9: Electrical stabilisation of a CdTe specimen during light-soaking at 1000 W/m² at a module temperature of 79.4°C ........................................................................................ 13

Fig. 10: Stabilisation of a CdTe specimen during light-soaking at 26.8°C module temperature and 1000 W/m² ............................................................................................................... 14

Fig. 11: Relative change of the nominal power of CIGS specimens during dark storage at various temperatures ...................................................................................................... 15

Fig. 12: Relative change of the nominal power of CdTe specimens during dark storage at various temperatures ...................................................................................................... 15

Fig. 13: Metastable evolution of the current-voltage characteristic measured for a CdTe module upon light exposure ............................................................................................ 16

Fig. 14: Open circuit voltage (Voc), fill factor (FF) and maximum power (Pmax) of a commercial CdTe module depending on light exposure time and electrical bias ................................ 17

Fig. 15: Pairs of modules were repeatedly characterised after stabilisation by light-soaking. Between measurements, modules were stored in the dark unbiased (left) and biased at Imp (right). ........................................................................................................................ 20

Fig. 16: Two modules were continuously measured upon outdoor exposure after unbiased dark storage (black points) and biased at I(Vmp) (red points). Pmax is shown normalised to the final power and corrected for module back temperature and irradiance. [11]. ............ 21

Fig. 17: Single and multi-junction PV devices compared to conventional voltage generators [3].23

Fig. 18: SR measurement setup, based on: (a) lock-in technique; (b) a filtered pulsed simulator. The shunt resistor may be replaced by a trans-impedance amplifier (TIA) [28].28

Fig. 19: Spectral response of different module types (CIS, a-Si, c-Si) [4]. .............................. 33

vii

Fig. 20: Modelled dependence of Isc on Air Mass of incident radiation, for single junction a-Si (blue circles) and crystalline silicon (red squares). The dark lines are linear fits to the modeled data points [6]. .................................................................................................. 34

Fig. 21: Left: Overview of the seasonal effects of temperature (εt), SWE (εswe), reflection (εr) and spectrum (εs) on a-Si performance. The effects are expressed as performance factors, the ratio of actual vs. rated performance due to each factor. The total performance factor is obtained by multiplication of the four individual performance factors [5]. Right: Seasonal changes of selected amorphous silicon devices [19]. ...................... 36

Fig. 22: Monthly average DC performance ratio for thin-film CIGS and CdTe systems over the period June 2006 – June 2010 in Nicosia, Cyprus [14]. ................................................... 36

Fig. 23: Meteorological data during monitoring period 2010-2012 at seven different sites. The upper figure shows the different distributions of daily irradiation sums, the lower figure shows the monthly average ambient temperatures measured at the sites. ...................... 39

Fig. 24: Average back-of-module temperature for irradiances G > 0 W/m² (line plots) and daily PR for Pmax (dot plots) for c-Si modules at five different sites during monitoring period May 2010 to Dec 2012 ............................................................................................................ 40

Fig. 25: Example of analysis approach with daily performance ratio PR(Pmax) (top/right) and PR(Isc) (top/left) of 6 different c-Si modules measured at different locations. Step 1: merge to normalised performance ratio norm. PR(Pmax)and Step 2: plot in dependence of daily irradiation (bottom/left) and of average daylight module temperature (bottom/right). ....... 41

Fig. 26: Average back-of-module temperature for irradiances G > 0 W/m² (line plots) and daily PR of Pmax (dot plots) for a-Si based modules measured in different locations during monitoring period May 2010 to Dec 2012. ....................................................................... 42

Fig. 27: Example of analysis approach with daily performance ratio PR(Pmax) (top/right) and PR(Isc) (top/left) of 7 different a-Si based silicon modules measured in different locations. Step 1: merge to normalised performance ratio norm PR(Pmax) and Step 2: plot in dependence of daily irradiation (bottom/left) and of average daylight module temperature (bottom/right). .................................................................................................................. 43

Fig. 28: Relative spectral response signal of different modules, illustrating significant differences within the same thin-film technology.............................................................. 45

Fig. 29: Angular characteristic of the best performing input optic: spectrally neutral but increasing cosine error above 40° ................................................................................... 46

Fig. 30: Angular characteristic of the worst performing input optic: wavelength dependent cosine error ..................................................................................................................... 46

Fig. 31: Solar spectral irradiance drift for four cloudless days in summer (red), spring (orange), all (yellow) and winter (blue), mounted facing south and tilted 35°, in Cologne ................ 48

Fig. 32: Characterization of a partly cloudy day (29.08.2010) and correlation of APE values with the appearance of clouds ......................................................................................... 48

Fig. 33: Seasonal variations of APE values and average APE value per month ..................... 49

Fig. 34: Dependence of measured Isc on spectral changes (T, G corrected, normalised to ISC, STC (=100%) at 1.65 eV) ................................................................................................ 50

Fig. 35: Average spectrum of the period 01.09.11–31.08.12 relative to AM1.5, area normalised50

Fig. 36: Comparison of horizontal simulated and measured irradiance for ABD test installation, year 2011. Top: unfiltered points (RMSE=80.2). Bottom: clear sky conditions (RMSE=31.8) .................................................................................................................. 53

viii

Fig. 37: Comparison of in-plane simulated and measured irradiance for ABD test installation, year 2011. Top: unfiltered points (RMSE=100.870). Bottom: clear sky conditions (RMSE=71.264) .............................................................................................................. 54

Fig. 38: Normalised efficiency against average wavelength, for clear sky conditions, angle of incidence less than 50°, sorted for shading (left) and filtered and corrected for standard temperature of 25°C (right). Different in-plane irradiance levels are displayed. ............... 57

List of Tables

Tab. 1: State-of-the-art efficiencies for various single- and multi-junction technologies, compared with the c-Si record device and with the III-V multi-junction record cell. All records refer to 1000 W/m2, AM1.5g spectrum and 25°C cell temperature [4-5]. ............. 24

Tab. 2: Parameters defined and used to perform common data analysis ............................... 39

1

1 Foreword

The International Energy Agency (IEA), founded in November 1974, is an autonomous body

within the framework of the Organization for Economic Co-operation and Development

(OECD) which carries out a comprehensive programme of energy co-operation among its

member countries. The European Union also participates in the work of the IEA. Collaboration

in research, development and demonstration of new technologies has been an important part

of the Agency’s Programme.

The IEA Photovoltaic Power Systems Programme (PVPS) is one of the collaborative R&D

Agreements established within the IEA. Since 1993, the PVPS participants have been

conducting a variety of joint projects in the application of photovoltaic conversion of solar

energy into electricity.

The mission of the IEA PVPS programme is: To enhance the international collaborative efforts

which facilitate the role of photovoltaic solar energy as a cornerstone in the transition to

sustainable energy systems.

The underlying assumption is that the market for PV systems is rapidly expanding to

significant penetrations in grid-connected markets in an increasing number of countries,

connected to both the distribution network and the central transmission network.

This strong market expansion requires the availability of and access to reliable information on

the performance and sustainability of PV systems, technical and design guidelines, planning

methods, financing, etc., to be shared with the various actors. In particular, the high

penetration of PV into main grids requires the development of new grid and PV inverter

management strategies, greater focus on solar forecasting and storage, as well as

investigations of the economic and technological impact on the whole energy system. New PV

business models need to be developed, as the decentralised character of photovoltaics shifts

the responsibility for energy generation more into the hands of private owners, municipalities,

cities and regions.

The overall programme is headed by an Executive Committee composed of representatives

from each participating country and organization, while the management of individual research

projects (Tasks) is the responsibility of Operating Agents. By late 2013, fourteen Tasks were

established within the PVPS programme, of which six are currently operational.

The overall objective of Task 13 is to improve the reliability of photovoltaic systems and

subsystems by collecting, analysing and disseminating information on their technical

performance and failures, providing a basis for their assessment, and developing practical

recommendations for sizing purposes.

The current members of the IEA PVPS Task 13 include:

Australia, Austria, Belgium, China, the European Photovoltaic Industry Association (EPIA),

France, Germany, Israel, Italy, Japan, Malaysia, the Netherlands, Norway, Spain, Sweden,

Switzerland, Turkey and the United States of America.

2

Performance characterization for thin-film PV modules requires special care because of the

complex nature of the devices compared to first-generation PV devices. This report addresses

the characterization of PV modules in solar simulators and in field conditions. We review the

previous work on best practices for pre-conditioning and indoor measurements and propose a

new method for stabilisation of modules without light exposure. Special considerations for

bifacial PV modules are also discussed. After summarizing previous studies on

characterization of field performance, we present a new, uniform analysis of international field

data from many contributors. Finally, we present case studies related to the effects of

spectrum on outdoor performance.

The editors of the document are Timothy J Silverman, National Renewable Energy Laboratory,

Golden, CO, USA, and Ulrike Jahn, TÜV Rheinland Energie und Umwelt GmbH, Cologne,

Germany (DEU).

The report expresses, as closely as possible, the international consensus of opinion of the

Task 13 experts on the subject at hand. Further information on the activities and results of the

Task can be found at: http://www.iea-pvps.org.

2 Executive Summary

Although thin-film photovoltaic (PV) modules have been in production for decades, the

characterization of their performance, both outdoors and under artificial light, remains a topic

of active research. This is because the field contains a diverse set of PV technologies, each of

which has physical differences from conventional crystalline silicon PV. These differences

range from different temperature coefficients to complex short-term or seasonal transients in

performance. This report summarises the nature of these special behaviours and

demonstrates best practices for handling them in the context of several case studies.

The first portion of the report deals with the performance of thin-film PV modules in solar

simulators. Achieving repeatable performance measurements is challenging, even under

artificial light. Stable, spectrally matched reference modules are generally unavailable, which

can lead to errors in the effective illumination level. Some technologies have high capacitance,

leading to problematic dependence of results on the duration of illumination and of the I-V

curve sweep. Modules must be stabilised to avoid the influence of metastabilities on the

results. This is normally done by light exposure or “light soaking” and the report proposes an

approach for improving upon simple light soaking by applying bias.

The report then covers the issues surrounding the measurement and analysis of outdoor

performance of thin-film PV modules. The widely varying spectral responses, temperature

coefficients and metastable behaviours of different thin-film technologies lead to special

challenges in outdoor performance analysis. This is illustrated by the presentation of a new

method of collecting and analyzing performance data from many international partners.

Specific issues of spectral performance analysis are then discussed, including measured-

spectrum and modeled-spectrum methods.

The target audience of the report is scientists and engineers who participate in the collection,

analysis and prediction of indoor and outdoor performance data. This group includes planners,

3

operators and manufacturers of PV power plants, participants in the standardization of

methods for performance measurement and workers in academe or at national laboratories.

3 Simulator Performance

3.1 Previous work

3.1.1 Introduction

G Friesen, University of Applied Sciences of Southern Switzerland (SUPSI)

A Louwen and WGJHM van Sark, Utrecht University, Copernicus Institute of Sustainable Development,

the Netherlands

Characterizing the performance of thin-film PV modules indoors is complicated by several

physical differences between thin-film and conventional crystalline silicon PV technology.

These differences can result in substantial scatter when the same module is measured in

different ways or at different laboratories. Endeavors that rely on the repeatability of indoor

performance measurements must be undertaken with these issues in mind.

The following contribution summarises a campaign of round-robin tests intended to

characterise and reduce the uncertainty in performance measurements from multiple

laboratories. This section is followed by a review of literature enumerating the various

phenomena that contribute to the difficulty in achieving repeatable indoor performance

measurements.

3.1.2 State of the art in indoor performance characterization of PV modules

with solar simulators

This chapter summarises results of previous international studies, which can be considered to

be representative of the state of the art of thin-film module characterization and its progress

compared to the better-investigated characterization of crystalline silicon modules. Most of the

work referenced here is from the European project IP PERFORMANCE (IP Contract 019718),

which ended in 2010.

One of the major objectives of the PERFORMANCE project was to improve the accuracy and

comparability of electrical power measurements of PV modules. Thus, standard test

conditions (STC) were considered, as well as all test conditions defined within the IEC 61215

[1] and IEC 61646 [2] module qualification standard: STC (1000 W/m², 25°C, AM1.5); different

module temperatures (800 W/m², 25-65°C, AM1.5), nominal operating cell temperature (800

W/m², NOCT, AM1.5) and low irradiance (200 W/m², 25°C, AM1.5).

New test equipment and procedures were developed within the project, to improve the

accuracy and to cover an even broader range of test conditions (100-1200 W/m2 and 15-75°C),

as requested by the upcoming IEC 61853 Energy Rating standard part 1 [3] and 2 [4].

The comparability of measurements was mainly assessed by the means of different

measurement intercomparison campaigns, so-called round-robin tests. All major European

test laboratories with broad experience in indoor and outdoor testing and ISO17025

4

accreditation participated in the round-robin tests. Two round-robin tests focused on crystalline

silicon modules and two on thin-film modules. The whole range of commercially available

products was considered. The execution of an initial round-robin with a follow-up round-robin

enabled the verification of the level of improvement achieved by the introduction of new test

procedures and equipment, all aiming to better meet the technology-specific requirements.

The round-robin results have been published at the major European conferences [5-7] and

merged into a “Best practice guideline for industry” [8].

Crystalline silicon round-robin results

Very good results have been achieved for all crystalline silicon modules tested within the

round-robin, with a comparability of Pmax of ±1-1.5% (see Fig. 1). For the majority of the

modules the spread was below ±1%. Only for high-efficiency modules, where special

measurement techniques are required to overcome capacitive effects [8], does the spread

reach ±1.5%. New measurement techniques have been developed more recently for high-

capacitance modules [9,10], and further round-robins are underway to assess the accuracy of

these new methods. The spread was significantly higher at test conditions different from STC

as shown in Fig. 2 and 3, but always within the limits of the declared measurement

uncertainties U[k=2]. All measurement uncertainties were calculated according to a common

approach agreed to within the PERFORMANCE project and published by Müllejans et al. [11],

which covers all relevant optical, electrical and measurement related factors. The increase of

the spread up to ±4% for low irradiance levels can be mostly explained by the increase in non-

uniformity of the solar simulators when introducing spectral neutral filters for the attenuation of

irradiance, whereas the spread of up to ±10% for the temperature coefficient of Pmax can be

mostly explained by the differences in test procedures and equipment available in the different

laboratories.

Fig. 1: Spread of Pmax at STC (deviation from average of six test laboratories) as measured with

different solar simulators and four different crystalline silicon module types

5

Fig. 2: Spread of Pmax at 200 W/m² (deviation from average of six test laboratories) as measured

with different solar simulators and four different crystalline silicon module types

Fig. 3: Spread of Pmax temperature coefficient measured at 800 W/m² (deviation from average of

six test laboratories) as measured with different solar simulators and four different crystalline

silicon module types.

Thin-film round-robin results

For thin-film modules the electrical characterization turned out to be much more difficult [6, 7]

and the round-robin was therefore limited to the characterization under STC. Electrical

instability of the material in combination with non-harmonised preconditioning techniques,

spectral mismatch problems caused by the unavailability of perfectly spectral matched

reference devices or spectrally tunable solar simulators, as well as the current limiting

behaviour of sub cells within multi-junction modules, resulted in a poorer comparability of

6

power measurements of thin-film modules. Fig. 4 gives an overview of the results obtained

with the final round-robin. Depending on the technology, a comparability of Pmax within 3-6%

was achieved for Pmax at STC, with the best result for crystalline silicon on glass (CSG) where

instability and spectral mismatch issues are less critical. A general improvement could be

demonstrated within the project from the first to the second thin-film round-robin by

harmonizing the stabilisation procedures. The following paragraphs highlight some

technology-specific problems that could not be solved within the PERFORMANCE project and

which will be treated in more detail throughout the whole report.

Fig. 4: Spread of Pmax at STC (deviation from average of six test laboratories) as measured with

different solar simulators and four different thin-film module technologies within the second

round-robin

For amorphous silicon the standard approach according to IEC 61646 was applied to stabilise

the module power, but with a stricter limit of ±1% for the second round-robin. It was observed

that the stabilisation within ±2% did not fully stabilise the modules, and the stabilisation

process continued during the execution of the round-robin, leading consequentially to a higher

spread. Thermal annealing was avoided by monitoring the module temperature during the

whole round-robin process and by not exposing the modules to temperatures above 50°C.

For copper indium (gallium) diselenide (CIS) and cadmium telluride (CdTe) a 1-hour outdoor

exposure under load at irradiance above 800 W/m² and temperature below 55°C was agreed

upon for pre-conditioning. A better comparability was achieved, but without being able to

guarantee full stability. Moreover, no harmonised procedure was suggested on how to deal

with the effects occurring during the timeframe from when the module is transported to the

solar simulator and the module is cooled down to 25°C.

The spectral mismatch error has been identified as a major issue for the characterization of all

thin-film technologies. An accurate measurement of spectral response and solar simulator

spectrum are of primary importance for spectral mismatch corrections or the estimation of the

7

spectral mismatch error when no spectral correction is applied. At the time of the project, only

one laboratory was able to measure the spectral response of complete modules and only of

single-junction devices. These measurements together with some spectral response curves

measured on the cell level have been distributed to all round-robin partners for the selection of

the best reference device and/or to perform spectral mismatch corrections. The comparability

between cell and module were very good, but the transferability of measurement results from

cell to module level has still to be demonstrated for all technologies. The spectra of the solar

simulators have all been measured with the same spectral radiometer, but without considering

variations over time caused by lamp aging or other issues. Later studies showed that thin-film

modules are particularly sensitive to spectral variations caused by the aging of the lamps or

other issues [12]. Also, the equipment for the spectral characterization of solar simulators is

under continuous investigation [13, 14]. New test equipment for the measurement of spectral

response curves of single and multi-junction modules has been introduced in the meantime by

most test laboratories [15,16,17], but too late to be included into the PERFORMANCE round-

robin campaigns.

3.1.3 Specific issues in thin-film modules

The round-robin results have highlighted some unresolved issues, which will be the main

focus of this report. Most of these issues are related to thin-film modules. A clear need for

further improvements was identified in the field of pre-conditioning of CIS and CdTe, the

testing of multi-junction devices, better spectral characterization of modules and solar

simulators, and the understating of seasonal variations within thin-film modules.

In the following sections, we give a short overview of the major issues that can occur with thin-

film technologies. Further information can be found in the “Best practice guideline for industry”

[4] published within the PERFORMANCE project.

Metastability and stabilisation procedures

Because of the metastable nature of the semiconductor materials, achieving precise power

measurements on thin-film PV modules often requires the application of a stabilisation

procedure prior to each measurement. According to the various physical mechanisms behind

the transient behaviours, many stabilisation techniques have been proposed. Some copper

indium gallium diselenide (CIGS) and CdTe manufacturers have helped test laboratories to

achieve higher power measurement accuracy by developing and sharing specific stabilisation

procedures for their own products.

Amorphous silicon

Single-junction amorphous silicon modules are well known to exhibit significant performance

degradation in the first few hundred hours of operation. This performance degradation leads to

a decrease of efficiency of ~10-30% [18], while a typical value is reported to be 16% [19]. After

several months of operation, degradations of even 30% to 40% were observed [19]. This

degradation is the result of a material degradation effect known as the Staebler-Wronski Effect

(SWE). With thermal annealing, this performance degradation can be partly recovered [19],

[18]. The strength of the effect is reported to be related to the device structure; devices with

thinner intrinsic layers exhibit less performance loss [18, 19], because of decreased

recombination of photocarriers [18].

8

The degradation necessitates preconditioning or stabilisation [19, 20]. Outdoor measurements

show rapid decrease in photocurrent in the first days of operation, and similar performance

degradation after 6 weeks of operation [15]. In IEC 61646 it is stated that stabilisation can be

assumed after 43 kWh/m2 of irradiation is applied and power output changes less than +/-2%.

Astawa et al [20] show that in order to perform accurate degradation measurements, loading

of the modules should be performed under realistic operating conditions (operation at

maximum power point). The practice of exposure to light, being indoor or outdoor, under open-

circuit conditions, leads to an overestimation of Voc degradation by 23% and an

underestimation of Isc degradation by 7.69% [20].

Cadmium telluride

The indoor characterization of CdTe modules is complicated by two factors: spectral mismatch

(discussed in the next section) and short-time light soaking effects [21]. Light soaking leads to

an initial performance improvement of ~4% [18], which necessitates preconditioning before

performing indoor performance characterization. The effect can be reversed through dark

storage [18].

Copper indium diselenide

CIS modules are known to exhibit “beneficial reversible metastability under light soaking”,

which necessitates preconditioning before testing device performance indoors [18]. In [21]

unconditioned CIS modules showed Pmax measured indoors to be 94% of that measured

outdoors. This light soaking effect can be reversed by annealing at 80°C in the dark [18].

Preconditioning times were found to be shorter when preconditioning was performed at higher

temperatures [18].

Spectral considerations for indoor measurements

As we have seen in previous sections, the effect of spectral variations on thin-film PV

performance can be quite pronounced. Especially for amorphous silicon (a-Si) and CdTe, this

means indoor characterization of modules should consider the incident spectrum of the solar

simulator, and correct for discrepancies between the simulator and solar irradiation at

standard test conditions.

Fig. 5 shows an overview of the effect of spectral mismatch on indoor characterization of

different PV parameters, for different thin-film technologies. Measurements for a-Si and CdTe

need correction for spectral mismatch because of the large discrepancy between the spectral

response of a-Si and CdTe on the one hand, and the c-Si reference cell on the other [21]. For

CIS, which has a very similar spectral response compared to c-Si, this effect is very much

smaller [21].

9

Fig. 5: Normalised PV parameters of single junction thin-film devices as measured (i) outdoors,

(ii) indoors without spectral mismatch correction, and (iii) indoors with spectral mismatch

correction. The variations are relative to the outdoor measurement [21].

Capacitive phenomena and I-V sweep time effects

Current-voltage characteristics are measured by sweeping voltage from slightly below zero to

above Voc, which takes a finite time. It has been found that I-V curves may differ as a function

of sweep time. Of the thin-film technologies, only a-Si (single, double and hybrid a-Si/µc-Si)

shows appreciable sweep time effects. For single-junction a-Si devices, sweep time effects

occur up to 10 ms [21]. For double-junction and hybrid a-Si, similar results were obtained, so

for these technologies, I-V measurements should be performed with sweep times >10 ms [21].

For CdTe and CIS, no sweep time effects were found. Fig. 6 shows the effect of sweep-time

for the different thin-film technologies mentioned.

Fig. 6: Normalised power (Pmax) of single-junction thin-film devices (left) and double-junction and

hybrid a-Si (right) module, measured as a function of I-V sweep-time. The dashed line indicates a

typical sweep-time used for the LAPSS solar simulator [21].

10

References for this section

[1] IEC 61215 (2005): Crystalline silicon terrestrial photovoltaic (PV) modules – Design qualification and type

approval.

[2] IEC 61646 (2008): Thin-film terrestrial photovoltaic (PV) modules – Design qualification and type approval.

[3] IEC 61853-1 edition 1.0. (2011): Photovoltaic (PV) Module Performance Testing and Energy Rating - Part 1:

Irradiance and Temperature Performance Measurements and Power Rating.

[4] IEC 61853-2 FDIS: Photovoltaic (PV) Module Performance Tasting and Energy Rating - Part 2: Spectral

Response, Incidence Angle and Module Operating Temperature Measurements.

[5] W. Herrmann, S. Mau, F. Fabero, T. Betts., N. van der Borg, K. Kiefer, G. Friesen, W. Zaaiman, Advanced

intercomparison testing of PV Modules in European test laboratories, Proc. 22nd EU PVSEC, Milan, Italy, (2007),

pp. 2506-2510.

[6] W. Herrmann, S. Zamini, F. Fabero, T.R. Betts, N.J.C.M. van der Borg, K. Kiefer, G. Friesen, W. Zaaiman,

Results of the European PERFORMANCE Project on Development of Measurement Techniques for Thin-Film PV

Modules, Proc. 23rd EU PVSEC, Valencia, Spain, (2008), pp. 2719 – 2722.

[7] W. Herrmann, S. Zamini, F. Fabero, T. Betts, N. van der Borg, K. Kiefer, G. Friesen, H. Müllejans, H.-D. Mohring,

M.A. Vázquez, D. Fraile Montoro, PV Module Output Power Characterization in Test Laboratories and in the PV

Industry - Results of the European PERFORMANCE Project, Proc. 25th EU PVSEC / 5th World PVSEC, Valencia,

Spain, (2010), pp. 3879 – 3883.

[8] E. Dunlop, F. Fabero, G. Friesen, W. Herrmann, J. Hohl-Ebinger, H.-D. Mohring, H. Müllejans, A. Virtuani,

SUPSI W. Warta, W. Zaaiman, S. Zamini, Guideline for PV power measurement in industry, EUR 24359 EN,

(2010).

[9] A. Virtuani, P Beljean, G. Rigamonti, G. Friesen, D. Chianese., Fast and accurate methods for the performance

testing of highly efficient c-Si Photovoltaic modules using a 10 MS single-pulse solar simulator and customised

voltage profiles, Meas. Sci. Technol. 23 (2012) 115604 pp. 1-8, doi:10.1088/0957-0233/23/11/115604

[10] C. Monokroussos, D. Etienne, K. Morita, C. Dreier, U. Therhaag, W. Herrmann, Accurate Power

Measurements of High Capacitance PV Modules with Short Pulse Simulators in a Single Flash, Proc. 27th EU

PVSEC, Frankfurt, Germany, (2012), pp. 3687 – 3692.

[11] H. Müllejans, W. Zaaiman, R. Galleano; Analysis and mitigation of measurement uncertainties in the

traceability chain for the calibration of photovoltaic devices; Measurement Science and Technology 20, (2009),

075101 pp. 1-12

[12] W. Herrmann, L. Rimmelspacher, Uncertainty of Solar Simulator Spectral Irradiance Data and Problems with

Spectral Match Classification, Proc. 27th EU PVSEC, (2012), pp. 3015 – 3021.

[13] P. Sperfeld, D. Dzafic, F. Plag, F. Haas, C. Kröber, H. Albert, S. Nevas, S. Pape, Usability of Compact Array

Spectroradiometers for the Traceable Classification of Pulsed Solar Simulators, Proc. 27th EU PVSEC, Frankfurt,

Germany, (2012), pp. 3060 – 3065.

[14] D. Polverini, R. Galleano; The use of fast spectroradiometers for pulsed solar simulator spectra measurements,

Proc. 37th IEEE PVSC, Seattle, WA, USA, (2011) pp. 003596-003600.

[15] M. Pravettoni, M. Nicola, G. Friesen, A Filtered Pulsed Solar Simulator for Spectral Response Measurements

of Multi-Junction Modules of Commercial Size, Proc. 27th EU PVSEC, Frankfurt, Germany, (2012), pp. 3209 –

3213.

[16] F.P. Baumgartner, D. Schär, S. Achtnich, Spectral Response Measurement of Tandem Modules, Proc. 27th

EU PVSEC, Frankfurt, Germany, (2012), pp. 3022 – 3026.

[17] Y. Tsuno, Y. Hishikawa and K. Kurosawa, A Method for Spectralour Response Measurements of Various PV

Modules, Proc. 23rd EU PVSEC, Valencia, Spain, (2008), pp. 2723 – 2727.

[18] M. Gostein and L. Dunn, Light soaking effects on photovoltaic modules: Overview and literature review, Proc.

37th IEEE PVSC, Seattle, WA, USA, (2011), pp. 3126–3131.

11

[19] M. Muñoz-García and O. Marin, Characterization of thin-film PV modules under standard test conditions:

Results of indoor and outdoor measurements and the effects of sunlight exposure, Solar Energy, vol. 86, no. 10,

(2012), pp. 3049–3056.

[20] K. Astawa, T. R. Betts, and R. Gottschalg, Effect of loading on long term performance of single junction

amorphous silicon modules, Solar Energy Materials and Solar Cells, vol. 95, (2011), pp. 119–122.

[21] A. Virtuani, Comparison of indoor and outdoor performance measurements of recent commercially available

solar modules, Progress in Photovoltaics: Research and Applications 19.1 (2011), pp. 11-20.

3.2 Stabilisation issues particular to thin-film PV modules

3.2.1 Introduction

Determining the STC power rating of a thin-film PV module is important for predicting outdoor

performance and for quantifying degradation caused by outdoor exposure or by accelerated

testing. This otherwise simple measurement is frustrated by the fact that most polycrystalline

thin-film PV modules exhibit some form of transient change in performance on time scales of a

few days or less. Modules must therefore be stabilised before their STC power rating can be

determined. The development of stabilisation methods that apply to all thin-film PV

technologies continues to be an active area of research because each technology, and in

some cases each specimen of a particular technology, behaves differently.

The following contribution is a report of new data showing the temperature dependence of the

light-soaking effect on a large population of full-size modules. This information helps to

illustrate the variety of transient responses that is possible in polycrystalline thin-film PV

modules. The section is followed by a recommended approach for further development of

procedures that do not rely solely on light exposure to achieve stabilisation.

3.2.2 Recent results in thin-film stabilisation

Markus Schweiger, TÜV Rheinland

Introduction

The physical processes of metastable behaviour of CIGS and CdTe modules are not

understood yet [1]. IEC 61646 [2] contains the only standardised stabilisation procedure for

thin-film PV-modules. This procedure does not take into account the technology-specific

characteristics of the different thin-film technologies.

The metastable behaviour of 22 specimens has been investigated. Three CIGS module types

from two different manufacturers and two CdTe module types from two different

manufacturers have been measured indoors. The change of the main module parameters Isc,

Voc and Pmax was monitored online while light-soaking and by regular re-measurements in a

pulsed solar simulator. The Pmax was recorded every 30 seconds and the entire I-V curve was

measured at 120-second intervals. The influence of the module temperature during the

stabilisation was investigated for 25°C, 50°C and 85°C. The irradiance during the stabilisation

process was 1000 W/m². The effect of small fluctuations of the module temperature and the

irradiance have been corrected according to IEC 60891 [3] procedure 2. After the stabilisation

program the modules were stored in the dark at 25°C with periodic I-V measurement to

monitor their de-stabilisation.

12

Stabilisation of CIGS specimens

CIGS modules stabilised at low temperatures (25ºC, 50°C) showed a rise of the nominal

power, compared to the initial value, of up to 8%. The changes occurred because of a rising fill

factor and a rising open circuit voltage, as shown in Fig. 7. The short circuit current remained

almost stable. Aging effects could be detected for some modules during stabilisation at higher

module temperatures (85°C), as shown in Fig. 8. The nominal power dropped down to -9% for

some module types. The irradiation of one 43 kWh/m2 cycle according to IEC 61646 appeared

to be sufficient.

Fig. 7: Electrical stabilisation of a CIGS specimen during light-soaking at 1000 W/m² and a

module temperature of 58.4°C

Fig. 8: Irreversible degradation of a CIGS specimen caused by light-soaking at 83.7°C module

temperature, online-monitored module parameters at 1000 W/m²

0.99

1.00

1.01

1.02

1.03

1.04

0:00 6:00 12:00 18:00 24:00 30:00 36:00 42:00

Rela

tavie

change t

o initia

l valu

e

time [h]

Pmpp Voc Isc

0.88

0.90

0.92

0.94

0.96

0.98

1.00

1.02

0:00 6:00 12:00 18:00 24:00 30:00

Rela

tavie

change t

o initia

l valu

e

time [h]

Pmpp Voc Isc

13

Stabilisation of CdTe specimens

Most CdTe modules showed a rise of the nominal power compared to the initial value of up to

10%. The changes mostly occurred because of a rising fill factor, as shown in Fig. 9. The

short-circuit current remains almost stable. In contrast to CIGS, the CdTe specimens showed

a strong dependency of the stabilisation result on the module temperature during the

stabilisation. The higher the module temperature was, the higher the resulting value of the

nominal power. Low temperatures (25°C) led to small positive or even negative changes of

nominal power, as shown in Fig. 10. A superposition of light-induced degradation and thermal

activation can be assumed. Tests with high module temperatures and without exposure to

light also showed a rise in the nominal power. The irradiation of one 43 kWh/m² cycle

according to IEC 61646 also seems to be sufficient for CdTe.

Fig. 9: Electrical stabilisation of a CdTe specimen during light-soaking at 1000 W/m² at a module

temperature of 79.4°C

0.96

0.97

0.98

0.99

1.00

1.01

1.02

1.03

0:00 6:00 12:00 18:00 24:00 30:00 36:00 42:00

Rela

tavie

change t

o initia

l valu

e

time [h]

Pmpp Voc Isc

14

Fig. 10: Stabilisation of a CdTe specimen during light-soaking at 26.8°C module temperature and

1000 W/m²

Relaxation during storage in the dark

The reversal of stabilisation was investigated by storing the specimens in the dark at 25°C for

several months and regularly performing re-measurements using the pulsed solar simulator.

For both technologies (CIGS and CdTe), the results show long relaxation times of several

weeks (Fig. 11 and 12). Some manufacturers state the destabilisation of their CIGS modules

occurs within seconds. For these fast transient effects, another measurement approach must

be chosen as described above.

0.94

0.95

0.96

0.97

0.98

0.99

1.00

1.01

1.02

0:00 3:00 6:00 9:00 12:00 15:00

Rela

tavie

change t

o initia

l valu

e

time [h]

Pmpp Uoc Isc

15

Fig. 11: Relative change of the nominal power of CIGS specimens during dark storage at various

temperatures

Fig. 12: Relative change of the nominal power of CdTe specimens during dark storage at various

temperatures

0.90

0.92

0.94

0.96

0.98

1.00

1.02

1.04

1.06

1.08

1.10

0 240 480 720 960 1200 1440 1680

Rela

tive c

hange t

o initia

l pow

er

valu

e

time [h]

Spec1_58.4°C Spec2_85.7°C Spec3_28.2°C

Spec4_64.4°C Spec5_55.9°C Spec6_23.4°C

0.9

0.92

0.94

0.96

0.98

1

1.02

1.04

1.06

1.08

1.1

1.12

0 240 480 720 960 1200 1440 1680

Rela

tive

ch

ang

e t

o in

itia

l p

ow

er

va

lue

time [h]

Spec1_79.4°C Spec2_26.7°C Spec3_44.3°C

Spec4_53.5°C Spec5_27.0°C

16

References for this section

[1] T. Eisenbarth: Identifikation von Defekten und Metastabilitäten in Cu(In,Ga)Se2-Dünnschichtsolarzellen, (2010),

Doctoral dissertation, Freie Universität Berlin.

[2] IEC 61646: Thin-film terrestrial photovoltaic (PV) modules – Design qualification and type approval, (2008-05).

[3] IEC 60891: Photovoltaic devices – Procedures for temperature and irradiance corrections to measured I-V

characteristics, (2009-12).

3.2.3 Improving upon thin-film PV stabilisation methods

Timothy J Silverman, National Renewable Energy Laboratory, United States Sabrina Novalin, Austrian Institute of Technology

Introduction

This chapter establishes the need to stabilise thin-film PV modules before measurement and

describes the shortcomings of the light-soaking method of stabilisation. Improvements of the

method, based on the use of bias in addition to or in place of illumination, are proposed. The

specific tests needed to develop these improvements are proposed and results from pilot tests

on CIGS modules are presented.

Background

The dominant thin-film photovoltaic materials, namely CdTe, CIGS and a-Si, all exhibit

transient changes in performance upon exposure to light [1-3]. These effects influence

current-voltage parameters and can result in either artificially high or low performance. The

metastable effects vary in magnitude and can be observed in cells and modules alike. In some

cases the effects might be close to measurement accuracy or even non-existent. However, in

other cases, the performance of the cell or module can vary more than 10% above an initially

measured value. Both the open-circuit voltage (Voc) and the fill factor (FF) can be affected by

these transients, and generally speaking, a partial or full relaxation to an originally measured

value can be obtained by dark storage at 25°-80° C within times widely varying among the

technologies (from hours to weeks). Fig. 13 and 14 show a time evolution of current-voltage

characteristics upon light exposure which was observed for a CdTe module. After dark storage,

the curves typically relax to the initial condition.

Fig. 13: Metastable evolution of the current-voltage characteristic measured for a CdTe module

upon light exposure

17

Fig. 14: Open circuit voltage (Voc), fill factor (FF) and maximum power (Pmax) of a commercial

CdTe module depending on light exposure time and electrical bias

Nevertheless, the metastable behaviour does interfere with performance measurements when

such measurements are made shortly after manufacture or after extended periods in darkness,

such as during shipment or accelerated testing. It is important for performance measurements

to reflect post-transient performance because this is a PV module’s normal operating state.

There is, thus, a need to “stabilise” modules, bringing them to their normal outdoor state, when

measuring their performance after manufacture or darkness.

Modules can exhibit a great variety of transient behaviours with diverse physical origins

operating at a wide range of time constants [1]. We focus on the ones most troublesome for

performance measurements: reversible transients with time constants on the order of minutes

or hours. Long-term transients are too easily confounded with degradation, and irreversible

transients are not of concern for repeated measurements of the same module.

The most direct method of stabilising thin-film PV modules is to expose them to light until their

performance no longer changes. A specific version of this approach is promulgated in the

“light-soaking” process in International Standard IEC 61646. The process entails repeated

exposure of a module to light in increments of ≥43 kWh/m2 until the module’s power value no

longer changes by more than 2% between measurements. The standard allows the 43

kWh/m2 exposure increments to occur over any length of time, containing any number of dark

periods. It does not specify how soon after exposure a performance measurement must be

made.

In CIGS and CdTe cells, bias can cause transient performance changes that are similar to

those caused by illumination. The effect was observed in CIS in the 1980s [6], and in the

1990s it was shown that placing CIS and CdTe cells in forward bias can reduce the transient

in Voc that occurs upon illumination [7]. The process was tested on full-sized CIGS and CdTe

modules in 2010. Although similar changes were observed between biased and light-exposed

modules, it was not possible to establish the equivalence of bias to light exposure [8].

Potential for improvements

The IEC 61646 light-soaking process was designed for amorphous silicon, which shows a

substantial degradation in response to light exposure, but a very slow recovery at room

18

temperature. The procedure was not designed for metastabilities that may relax quickly at

room temperature. For example, CIGS modules can return to their dark state within several

hours [9], suggesting that nighttime periods occurring during an exposure can cause some of

the light-soaking effect to be lost. Bringing modules to indoor measurement equipment from

outdoors can introduce a substantial delay that is usually not measured or controlled. This

exposure-measurement delay has been found to affect the measurement by allowing some of

the light-soaking effect to be lost [10]. This effect taints comparisons between modules in the

same batch and between different measurements of the same module. Whereas the IEC

61646 process works well for amorphous silicon, a different procedure should be developed

for CIGS and CdTe modules that show faster relaxation after light exposure.

For CIGS and CdTe modules, there is evidence that the equivalence of light and forward-bias

can be used to improve the stabilisation process. The main advantage offered by bias

stabilisation is that of improved control. Bias stabilisation is not subject to change due to

weather or day length. The exposure-measurement delay can be easily controlled and made

very short. There might also be cost and speed advantages: an outdoor light-soaking setup or

a continuous solar simulator might be replaced with an inexpensive DC power supply that

could be operated continuously, eliminating lost time because of poor weather. Bias conditions

might be identified that allow stabilisation to proceed much more rapidly than with light-

soaking.

Proposal for development of new stabilisation methods

The objectives of developing new stabilisation methods are to make the stabilisation process

better controlled, less expensive and faster than the existing light-soaking method. A new

method must also produce equivalent or better results for a particular technology or for a

specific module type. There is sufficient evidence to proceed with the investigation of two

approaches for improving the light-soaking method with bias stabilisation: elimination of the

dependence of performance results on the exposure-measurement delay and elimination of

light exposure from the stabilisation process entirely.

In this section, we describe experiments undertaken on CIGS modules to aid the development

of improved stabilisation methods. Until sufficient data are available to develop a standard test

method, these experiments can be used to establish the efficacy of a proposed method on a

specific module type.

Stabilisation during the exposure-measurement delay

We propose to determine whether placing a PV module in forward-bias during the delay

between light exposure and characterization eliminates the sensitivity of the performance

measurement to the length of the delay. To be successful, this approach must be equivalent to

ordinary light soaking followed by immediate I-V curve measurement.

The test makes use of a set of matching modules. All modules are exposed to light outdoors

or in a solar simulator until their performance no longer changes with time. Immediately upon

removal from the light source, an I-V curve is collected in a flash solar simulator. With the

exception of one control module, the modules are then immediately placed in forward-bias. A

different constant-current bias condition may be used on each non-control module. Constant

current is used so that moderate changes in temperature do not cause large changes in the

19

bias condition. I-V curves are collected at increasing time intervals until the exposure-

measurement delay is several days.

The performance of the control module, which is stored in darkness between post-exposure

measurements, is expected to gradually change throughout the experiment. If it does not, the

module type under test does not require special treatment during the exposure-measurement

delay.

One or more of the modules placed in forward-bias during the delay are expected to show no

change in performance. Some modules may be damaged by the application of excessive

current; this test may be able to distinguish the threshold below which there is no risk of

damage. We suggest use of the lowest-current condition that maintains the module in its

stabilised state during the delay. Preliminary results are shown in a later section from a pilot

experiment on five types of CIGS modules.

Stabilisation without light exposure

Another proposed test is to determine the efficacy of forward-bias stabilisation with no light

exposure at all. The objective is to establish a process that is better controlled, less expensive

and faster than the light-soaking process. The test can also be used to determine the

conditions under which forward bias is equivalent to light exposure.

The test makes use of a set of matching modules that have been stored in darkness for

several weeks. An I-V curve is collected from each module at STC. With the exception of one

control module, the modules are then placed in forward-bias at several different constant-

current bias conditions. The voltage and temperature of the biased modules are continuously

monitored. I-V curves are collected at increasing time intervals and the test proceeds when

consecutive I-V curves are sensibly identical. All of the modules are then exposed to light and

their Pmax or Voc continuously monitored.

The performance of the control module is expected not to change during the indoor portion of

the test, then to exhibit its usual transient behaviour on exposure to light. If its performance

does not change outdoors, no stabilisation is needed for this module type. If its performance

changes during the indoor portion of the test, it may not have been in the dark state at the

start of the test.

One or more of the remaining modules are expected to change in performance during the

indoor portion of the test, then show no further change when deployed. The indoor

performance measurements will reveal when the modules became fully stabilised. The

temperature-corrected voltage measurements collected during bias can be used to establish a

simple stopping criterion for achieving stability. If various indoor bias conditions are tested, the

lowest-current condition that eliminates the transient upon light exposure can be identified.

Preliminary results from this method are shown in a later section.

CIGS pilot tests and results

We performed pilot tests on CIGS modules based on both of the preceding proposals. The

results show that further investigation of these methods is warranted.

20

Stabilisation during the exposure-measurement delay

To test exposure-measurement stabilisation, we used pairs of CIGS modules of five types [10].

Each type was tested at a different time. After extended storage in darkness, both modules

were exposed outdoors for several days. The modules were then brought indoors and within

10 minutes, an initial I-V curve was measured at STC. One module was then immediately

placed in forward-bias at Imp (forward-biased dark storage) and the other was left at open-

circuit (unbiased dark storage). I-V curves were collected from both modules at increasing

time intervals up to 96 h.

With most module types, performance declined as expected during unbiased dark storage, as

shown in Fig. 15. With the exception of CIGS D, the unbiased modules declined in Pmax by

≥1% within 1 hour and continued declining for the entire length of the experiment.

Fig. 15: Pairs of modules were repeatedly characterised after stabilisation by light-soaking.

Between measurements, modules were stored in the dark unbiased (left) and biased at Imp (right).

The forward-biased modules showed much smaller performance changes. With the exception

of CIGS A, the forward-biased modules produced the same Pmax, within 1% of the first

measurement made indoors. CIGS A is thought to have shown anomalous behaviour either

because its fast transient caused a decline in Pmax during the ≤600 s exposure-measurement

delay or because Imp forward-bias caused an artificial enhancement in performance during the

test.

The pilot data show that in many cases, placing a module in forward-bias at Imp immediately

upon removal from light exposure can eliminate the sensitivity of indoor performance

measurements to the length of the delay between exposure and measurement. The

exceptions are a module that showed no transient behaviour at all and one with such fast

behaviour that the test was not conclusive. These results are sufficiently promising to continue

testing this method with different module types, including CdTe modules, and with various bias

conditions. These steps will enable the design of the ideal method for making measurements

on modules with transient behaviour.

Stabilisation without light exposure

To determine the feasibility of bringing a module to its light-soaked performance level using

bias alone, we used a pair of matching CIGS modules. Both modules were stored in darkness

for several weeks. The test module was then placed in forward-bias at I(Vmp) and the control

21

module was left at open-circuit for several days. Both modules were then simultaneously

deployed outdoors in sunny conditions, biased at the maximum power point.

The maximum power was monitored during outdoor exposure for several days. The results

are plotted in Fig. 16. For the module type under test, stabilisation was achieved after <2

kWh/m2 of exposure. The control module’s Pmax increased by 15% in the first day of exposure;

the test module’s Pmax decreased by 3%. The bias stabilisation appears to have given the test

module an artificial performance increase of 3%.

Fig. 16: Two modules were continuously measured upon outdoor exposure after unbiased dark

storage (black points) and biased at I(Vmp) (red points). Pmax is shown normalised to the final

power and corrected for module back temperature and irradiance. [11].

The pilot data show that bias stabilisation with no light exposure at all is sufficiently promising

to continue investigating the method. Further analysis can reveal what combinations of bias

condition and duration can achieve stabilisation without producing an artificial performance

increase. It can also be determined whether the test module’s entire I-V curve, not just its Pmax

value, is brought to the same state as if it were stabilised with the light-soaking process. Tests

on multiple module types can show under what conditions the method is effective and can aid

in developing a stabilisation process that works for every module type or a set of stabilisation

processes for individual application to different module types.

Conclusion

Pre-measurement stabilisation of thin-film PV modules has shortcomings that may be

addressed by techniques involving the use of forward bias in addition to or in place of light

exposure. We have suggested tests and provided promising pilot data to aid in the

development of improved stabilisation methods. Stakeholders in thin-film PV characterization

are encouraged to participate in this development so that an improved standard method can

be established.

References for this section

[1] M. Gostein, M. and L. Dunn, Light soaking effects on photovoltaic modules: Overview and literature review, Proc.

37th IEEE PVSC, Seattle, WA, USA, (2011), pp. 003126-003131.

[2] S. Siebentritt, M. Igalson, C. Persson, S. Lany, The electronic structure of chalcopyrites - bands, point defects

and grain boundaries, Prog. Photovolt: Res. Appl. 18 (2010), pp. 390–410.

[3] J. A. del Cueto and B. von Roedern, Long-term transient and metastable effects in cadmium telluride

photovoltaic modules, Prog. Photovolt: Res. Appl. 14 (2006), pp. 615-628.

[4] A. Mittal, Master Thesis, Short Term Metastable Effects in the Amorphous Silicon Solar Modules, University of

Oldenburg, Germany, (2012).

[5] N. Taylor (ed.), Guidelines for PV Power Measurement in Industry, JRC Scientific and Technical Report EUR

24359 EN, doi:10.2788/90247, (2010).

22

[6] M.N. Ruberto, and A. Rothwarf, “Time-dependent open-circuit voltage in CuInSe2/CdS solar cells: Theory and

experiment,” J. Appl. Phys. 61 (9), (1987), pp. 4662-4669.

[7] R.A. Sasala, and J.R. Sites, “Time dependent voltage in CuInSe2 and CdTe solar cells,” Proc. 23rd PVSC

Louisville, KY, USA, (1993), pp. 543-548.

[8] J.A. del Cueto, C.A. Deline, S.R. Rummel and A. Anderberg, “Progress toward a stabilisation and

preconditioning protocol for polycrystalline thin-film photovoltaic modules,” Proc. 35th PVSC, Honolulu, HI, USA

(2010), pp. 002423-002428.

[9] L. Dunn. and M. Gostein, “Light soaking measurements of commercially available CIGS PV modules,” Proc.

38th IEEE PVSC, Austin, TX, USA, (2012), pp. 001260-001265.

[10] C.A. Deline, A. Stokes, T.J. Silverman, S. Rummel, D. Jordan and S. Kurtz, “Electrical bias as an alternate

method for reproducible measurement of CIGS photovoltaic modules,” Proc. SPIE 8472, San Diego, California,

USA, (2012), pp. 84720G-84720G.

[11] B. Marion, "Comparison of predictive models for photovoltaic module performance." Proc. 33rd IEEE PVSC,

San Diego, CA, USA, (2008), pp. 1-6.

3.3 Performance characterization of thin-film multi-junction PV cells

and modules

Mauro Pravettoni, University of Applied Sciences of Southern Switzerland (SUPSI)

Marco Apolloni, TEL Solar, Switzerland

3.3.1 Introduction

The aim of this section is to examine current-voltage characterization of thin-film multi-junction

cells and modules. In the absence of an international standard, which is under discussion at

the time of this writing, the authors and contributors have tried to put together their know-how

and the available national standards to draft a review of the basic principles. Literature works

that have treated all challenging aspects to date are examined, including very recent ones.

First, multi-junction photovoltaic devices are briefly introduced, including two challenging

aspects: current-limitation and spectral mismatch. Then the specific problems that may be

present in current-voltage characterization are addressed: choice of reference devices;

requirements for solar simulators; the challenges in outdoor characterization and in spectral

responsivity; and finally, the procedure for spectral mismatch correction according to the

laboratory practice, with the definitions of the matching factor Zi and of the current balance

Balij and their effect on the electrical parameters.

3.3.2 Theory and definitions

Multi-junction cells and modules

In a multi-junction cell, two or more PV junctions with decreasing bandgaps are stacked in optical and electrical series on top of each other. The junction with the widest bandgap faces the incoming irradiance and the one with the smallest bandgap is located at the bottom of the series, so that lower energy photons are able to pass through the top junctions and be absorbed by the bottom one. As a result, the total current is reduced, but the thermal energy loss of the high-energy photons is also reduced and the voltage is increased, with an overall positive effect on cell efficiency (see Fig. 17, where single- and multi-junction PV devices are compared to conventional batteries). A comprehensive work on multi-junction devices within third generation PV is included in the book by M. A. Green [1]; further information on thin-film multi-junction devices may also be found in Ref. [2].

23

(a)

(b)

Fig. 17: Single and multi-junction PV devices compared to conventional voltage generators [3].

A well-known configuration of a multi-junction cell is the two-terminal monolithic: the junctions

are grown on the top of each other and the electrical interconnection is provided by

appropriate tunnel-junctions between each pair of sub-cells. The device is therefore two-

terminal and electrical connection is possible to the total series of sub-cells. This technique is

used to build multi-junction cells from the III-V group (InGaP/GaAs, InGaP/GaAs/Ge and so

on), which because of their high manufacturing cost are usually only considered for aerospace

applications and in concentrator systems.

Multi-junction thin-film cells based on amorphous silicon (hydrogen alloy, a-Si:H) have also

been engineered, via inexpensive chemical vapour deposition (CVD) of the junctions, with

each pair separated by an intermediate layer reflector of Zinc oxide (ZnO) or Silicon oxide

(SiOx) that also enhance light trapping.

To date, there are multi-junction commercial modules on the market made of several

combinations of a-Si and micro-crystalline (mc-Si, also referred to as nano-crystalline, nc-Si)

junctions, such as: two-junction modules a-Si:H/a-Si:Ge and a-Si:H/mc-Si; three-junction

modules a-Si:H/a-Si:H/a-Si:Ge, and a-Si:H/mc-Si/mc-Si. Also “hybrid” structures, depositing a-

Si on mono-crystalline Silicon (c-Si) to form a-Si:H/c-Si has recently been studied and

developed. Tab. 1 shows the state of the art of various single-, two- and three-junction

structures, compared with c-Si and the overall record efficiency non-concentrator cell to date

(a 38.8% InGaP/GaAs/InGaAs cell): all values were measured at 1000 W/m2, AM1.5g

spectrum and 25°C.

24

Tab. 1: State-of-the-art efficiencies for various single- and multi-junction technologies,

compared with the c-Si record device and with the III-V multi-junction record cell. All records

refer to 1000 W/m2, AM1.5g spectrum and 25°C cell temperature [4-5].

Technology Junctions Wavelength

range [nm]

Short-circuit

current

density

[mA/cm2]

Open-

circuit

voltage

[V]

Cell

efficiency

[%]

Year

a-Si:H 1 300-800 16.75 0.886 10.1 2009

mc-Si 1 300-1200 26.55 0.549 10.7 2012

CdTe 1 300-900 28.59 0.857 19.6 2013

a-Si:H/a-Si:Ge 2 300-900 11.7 1.621 12.5 1992

a-Si:H/c-Si 2 300-1200 12.93 1.365 12.3 2011

a-Si:H/a-Si:H/a-Si:Ge 3 300-900 7.72 2.375 13.5 1996

a-Si:H/c-Si/c-Si 3 300-1200 9.52 1.963 13.4 2012

InGaP/GaAs/InGaAs 3 300-1300 14.27 3.065 38.8* 2013

c-Si (reference) 1 300-1200 42.7 0.706 25.0 1999

*http://www.pv-tech.org/news/spectrolab_surpasses_previous_own_world_record_for_cell_efficiency

Current limitation and spectral mismatch

Because junctions are electrically connected in a series, multi-junction cells and modules face current-

limitation by the sub-cell that generates the smallest photocurrent. Junction thickness and intermediate-

layer reflectors of multi-junction devices can be engineered and optimised so that each junction delivers

the same amount of photocurrent, but this is obviously dependent on the spectral irradiance of the

incoming light. Furthermore, if a device is designed to have current-matched junctions under a

reference spectrum (e.g., the standard AM1.5g) it may face strong spectral mismatches in operating

conditions under natural variations of the spectral irradiance during the day.

3.3.3 Basics of measurement methods

Current-voltage characterization: reference devices

Because of the multi-junction structure, the choice of reference devices for the detection of

total irradiance in current-voltage characterization is particularly challenging and may be

discussed under two options:

1. broadband reference cell (typically a calibrated c-Si reference cell): this option allows

the reference to detect the spectral irradiance over the widest possible wavelength

band, where the testing multi-junction device responds;

2. multiple component reference cells (typically calibrated, filtered c-Si reference cells [6],

while component cells based on thin-film materials are discouraged for stability

reasons): this option foresees the usage of a component cell for each junction, in order

to detect the fraction of total irradiance that is absorbed by each of the multiple

junctions.

The first option may be a good one for open-circuit voltage measurement, because all

junctions contribute equally to this parameter due to the series electrical connection [7]; but

the spectral mismatch discussed previously usually results in measurements of the short-

circuit current affected by strong uncertainties. From this point of view, component cells

guarantee that the current-limiting junction receives the appropriate fraction of total irradiance,

so that the measured short-circuit is correct.

25

The choice of the most appropriate reference cell among variously filtered samples can be

done, as suggested first in Ref. [8], by selecting the reference cell that shows the same

response (or minimal variations) to tiny spectral variations as the testing device. In fact, the

use of a perfectly matched reference cell bypasses the calculation of spectral mismatch

correction via spectral responsivity measurements, as specified in the following sections 3.4

and 3.5, and may be a practical tool for those labs which cannot perform such measurements.

Because open-circuit voltage measurement is subject to other major uncertainty contributions

(thermal and transient effects, above all), using component cells is definitely preferred in the

experimental practice; nevertheless, for solar simulators with spectral irradiance that is poorly

matched to the standard spectrum, the effect of spectral mismatch on the open-circuit voltage

with component reference devices should not be ignored. Furthermore, filters need to be

checked carefully and periodically to avoid ageing and subsequent need of recalibration. And,

in principle, laboratories involved in the characterization of many different technologies would

need a large number of component cells, depending on the growing demand and interest in

exotic multi-junction structures, which could be prohibitive. In these cases, even with the best-

matched component cell, a spectral mismatch correction is required, as discussed in a

dedicated section that follows.

All typical component reference cells show a specific angle of incidence (AOI) behaviour that

might be quite different from the regular cosine-behaviour. In this case, the characteristics

must be determined by goniometer measurements and need to be incorporated in the

measurement uncertainty calculation.

Indoor characterization: requirements for solar simulators

The previous section highlights a key factor in the current-voltage characterization of multi-

junction devices: the spectral mismatch between the testing light source and the reference

spectrum. This mismatch can easily give rise to huge uncertainties, even with Class A solar

simulators in spectral irradiance, i.e., within ±25% spectral match between the simulator’s

spectral irradiance and the reference spectrum [9] in the following six wavelength bands: 400-

500, 500-600, 600-700, 700-800, 800-900, 900-1100 nm, according to IEC 60904-9 [10].

The spectral irradiance matching requirement was introduced when multi-junction devices

were not yet developed and is still a valid requirement for first-generation c-Si cells and

modules. Discussion is ongoing whether to extend the short wavelength bands to 300 nm for

a-Si based devices that respond mainly in the first two of the six wavelength bands above; or

even up to 1800 nm, for some germanium-based CPV devices.

Another point under discussion in the PV community is the possibility of introducing a Class

A+ requirement, setting a spectral match limit to ±12.5%, as has already been suggested by

researchers at TÜV Rheinland, Germany [11]. Such a new limit may help to better control the

spectral mismatch correction, as specified in section 3.5.

While the American standard for solar simulator requirements ASTM E 927-10 [12] gives the

same requirements as the International one, the Japanese standard JIS C 8942 (for which

Class A spectral irradiance is reported as Class MA with further restrictions) defines a superior

Class MS, for simulators having spectral matching within ±5% [13]. It has to be noted that not

only is the matching requirement reduced from ± 25% to ± 5% for Class MS, but also the

26

range is extended down to 350 nm and, instead of 6 bands, 15 equidistant bands of 50 nm are

defined, which introduces a much stronger requirement than what is defined by either IEC or

ASTM.

Spectrally adjustable solar simulators based on multi-lamp systems are a useful option that

enables measuring variations of the electrical parameters of the testing module with spectral

variations. Also, it may be a very useful option to meet in principle any of the criteria specified

in section 3.5 to minimise the spectral mismatch. With a single lamp simulator the intensity of

light may be varied in a controlled way, resulting in spectral variations that may be useful for

understanding the multi-junction cell's response to changing spectra [14]. This variation is

regularly done by changing the current flowing through the lamps or flash tubes together with

varying the distance between the source and the testing device to reach optimal balancing

conditions.

Above all these topics, the final issue is related to the measurement of spectral irradiance,

which is also challenging, especially for pulsed solar simulators. Spectrometers with resolution

of 0.5 to 10 nm are available on the market and should be used to measure the spectral

irradiance: pulsed solar simulators require very fast-responding instruments, with integration

times between 1 and 10 ms. The integration time should be set to allow detection of any

variation of spectral irradiance within the duration of data acquisition.

Other important points to be addressed are the transient effects that may result from using

short-pulse solar simulators [15]. In fact, certain devices require measurements lasting

between 10 and 100 ms or more. Stationary measurements at fixed voltages (the so-called

"multiflash procedures") may be introduced to try avoiding these transient effects [16].

Outdoor characterization

Outdoor measurements represent a possible alternative to indoor ones, when the poor

spectral irradiance of a solar simulator leads to a large spectral mismatch. In this case,

measurements should be performed according to the following requirements:

AM1.5g spectral irradiance: it can be reached in clear-sky conditions on a tracker and

in certain times of the day throughout the year, depending on the geographical location

and the period of the year. Various tools may be found online, calculating the air mass

as a function of the geographical coordinates and day of the year.

Cell temperature must be kept close to 25°C over the duration of data acquisition.

Temperature coefficients may be determined experimentally [17].

Broadband reference cells may generally be used in outdoor measurements, because

of expected smaller spectral mismatch than in the indoor case; nevertheless, the real

spectral irradiance can show non-negligible differences with respect to the standard

AM1.5g spectrum. Therefore, the spectral irradiance should be measured and the

spectral mismatch should be evaluated as in Ref. [18].

Spectral responsivity

A standard procedure for the spectral responsivity (SR) measurement of multi-junction PV

cells and modules is ongoing, based on the standard SR measurement procedure for single-

junction devices [19]. With respect to the latter, the following additional requirements need to

be followed:

27

supplementary "coloured" bias light: Bias light is used in single-junction SR

measurements in order to put the test device in operating conditions or at least in its

range of linear response, when the monochromatic source is not intense enough to do

so. In addition, with multi-junction PV devices, bias light is used to avoid current

limitation by the junctions that are not being measured; this approach was proposed in

the late 1980s and is well-known in the literature [20-21].

forward bias voltage: With supplementary coloured bias light, the production of

current from testing a multi-junction device is strongly unbalanced, driving the

junction(s) not being tested into forward bias, while the testing junction is reverse-

biased. This may lead to an incorrect measurement of the device’s SR, in the presence

of shunting [22] and should be corrected by applying an appropriate forward bias

voltage to the test device.

luminescent coupling: With very high-quality materials a junction in forward bias also

experiences radiative coupling, and thus a junction at the top in forward bias, caused

by colour bias, may re-radiate light to the testing junction below [23-24]. As a

consequence, while measuring middle and bottom junctions some non-zero SR may

be measured in the wavelength band of the junction higher in the stack. This

contribution does not come from the SR of the testing junction, but as a side effect of

luminescent coupling from the junction right above. This effect sometimes increases

with increasing intensity of coloured bias light and needs to be treated as a different

effect from current-leakage caused by shunting. Recently, certain correction

procedures have been suggested in Ref. [25]. This effect is typically a minor one in

thin-film modules due to their poor material quality, while it is a dominant one in the

high-efficiency III-V multi-junction cells for CPV and space applications.

SR measurements on the testing device may be problematic with large-area modules. It is not

trivial to engineer continuous and chopped monochromatic light of commercial module size;

the lock-in technique may thus be used while measuring one cell at a time, appropriately light-

biasing the entire module and adding the appropriate forward bias (see Ref. [26]). Another

option is to use lenses or mirrors to drive and zoom a monochromatic small-size beam to the

larger target area [27].

Filtered pulsed simulators are a different option, not involving a lock-in technique, using an

instrument that may be already available in PV testing labs, but requiring expensive bandpass

filters to cover the full size of the lamp [28]: coloured bias light may again be superimposed on

a pulsed monochromatic beam of lower intensities, while additional bias voltage is applied

directly to the entire module as mentioned previously, when needed. Fig. 18 shows a

schematic of the lock-in technique setup and of the filtered pulsed simulator setup.

28

(a) (b)

Fig. 18: SR measurement setup, based on: (a) lock-in technique; (b) a filtered pulsed simulator.

The shunt resistor may be replaced by a trans-impedance amplifier (TIA) [28].

Spectral mismatch correction

A spectral mismatch correction factor M should be calculated according to IEC 60904-7 [29]

for each junction. The factor Mi corresponding to the i-th junction, being current-limiting under

given testing spectral conditions, should then be applied to the measured short-circuit current

to give the value corrected to AM1.5g. But depending on the choice of the reference detector

(see section 3.1) and of the spectral match of the solar simulator in use, Mi may show

deviations up to 10% from unity, with huge uncertainties in the final value of short-circuit

current. To limit that, it is recommended that researchers use component reference detectors

and adjust the spectral irradiance of the solar simulator to lower all Mi deviations from unity

below 3%. A limit of 1% would be optimal, but may require a lot of time and effort in spectral

adjustment and may not be achievable by all laboratories. Measurement uncertainty should

take the Mi into account.

The procedure for spectral tuning (when component reference cells are used and a spectrally

adjustable solar simulator is in use) is well described in the ASTM standard E2236-10 [30] and

is reported here for simplicity:

1. Adjust the total irradiance of the spectrally adjustable solar simulator to a level close to

the desired reporting conditions (1000 W/m2 for measurements at STC) using a

reference cell or previous experience (this initial irradiance level is not critical);

2. Measure the spectral irradiance of the spectrally adjustable solar simulator with a

calibrated spectroradiometer;

3. Calculate the spectral mismatch factor Mi for each junction, using the SR of the i-th

component reference cell;

4. Measure the short-circuit current of each component reference cell under the spectrally

adjustable solar simulator (IRi, with i = 1,...,N, being N the number of component cells,

equal to the number of junctions);

5. Calculate the matching factor Zi for the i-th junction according to

Z i=1

Mi

I sc,i,cal

IRi

,

29

where Isc,i,cal is the calibration value of the i-th component reference cell at the

reference total spectral irradiance (1000 W/m2 at STC).

6. If the matching factor for each component reference cell is within 3%, the spectrally

adjustable solar simulator is adjusted to within reasonable limits. If not, increase the

spectral content in the appropriate band and repeat steps 2 to 6 until the requirement

is met.

A similar method is provided also from the Japanese standard [31]. Zi and Mi values should be

provided in the measurement report, together with the related uncertainties.

Current balance and its effect on the electrical parameters

A further parameter that should be provided in the measurement report is the current balance

between the i-th and the j-th junction, which is defined as the ratio

where Ji is the density of photogenerated current by the i-th junction, having spectral

responsivity SRi() and being G() the testing spectral irradiance. Balij should be calculated for

both the incident spectral irradiance and for the reference spectrum and for all pairs of

junctions. While the matching factor Zi gives an indication of the spectral match of the light

source in use (with respect to the standard reference), the current balance Balij gives an

indication of the current match between junctions under the chosen light source.

Of course, Balij calculated for the testing light source should be close to the value calculated

for the standard reference, being a consequence of good spectral match. Though at the time

of this writing there is still no standard target value for it, in line with the requirements of

section 3.5 it is recommended that Balij under the test spectrum agrees to within at least ±3%

with respect to the reference spectrum (optimally ±1%, as in the standard for space

applications [32]).

All electrical parameters (open-circuit voltage, short-circuit current, fill factor and maximum

power) are affected by Balij. A plot showing their dependency over current balance should

therefore be reported, as in Ref. [33]. A recent survey between different laboratories showed

deviations in the maximum power within ±5-8% [34].

3.3.4 Acknowledgments

The authors acknowledge the contributions of Daniela Dirnberger and Holger Seifert at

Fraunhofer ISE, Germany; Sarah Kurtz and Keith Emery at the National Renewable Energy

Laboratory, United States; Werner Herrmann at TÜV Rheinland, Germany; Yoshihiro

Hishikawa at the National Institute of Advanced Industrial Science and Technology (AIST),

Japan; Harald Müllejans and Elena Salis at the European Commission, Joint Research Centre,

Institute for Energy and Transport, European Solar Test Installation, Italy; Alessandro Virtuani

and Gabi Friesen at the University of Applied Sciences and Arts of Southern Switzerland

(SUPSI), Switzerland; Johannes Meier at TEL Solar-Lab, Switzerland; Kerstin Keller at TEL

Solar, Switzerland and Tom Betts at Loughborough University, United Kingdom.

30

References for this section

[1] M. A. Green, Third generation photovoltaics. Advanced solar energy conversion. Springer-Verlag, Berlin, (2003).

[2] VV. AA., Thin-film silicon solar cells, edited by Arvind Shah, EPFL Press, Lausanne, (2010).

[3] M. Pravettoni, New Techniques for the Measurement of Second and Third Generation Photovoltaics, Imperial

College London, (2011).

[4] M. A. Green, K. Emery, Y. Hishikawa, W. Warta and E. D. Dunlop, Solar cell efficiency tables (version 42), Prog.

Photovolt: Res. Appl. 21, (2013), pp. 827-837.

[5] M. A. Green, K. Emery, K. Bucher, D. L. King and S. Igari, Solar Cell Efficiency Tables (Version 9), Prog.

Photovolt: Res. Appl. 5, (1997), pp. 51-54.

[6] J. Meier, R. Adelhelm, J. Hötzel, D. Romang, S. Bengali and U. Kroll, Reference cells in WPVS Design for

precise Micromorph PV power measurements, 25th

EU PVSEC / 5th

WCPEC, Valencia, (2010), pp. 2728-2734.

[7] M. Pravettoni, A. Virtuani, K. Keller, M. Apolloni and H. Müllejans, Spectral Mismatch Effect to the Open-circuit

Voltage in the Indoor Characterization of Multi-junction Thin-film Photovoltaic Modules, Proc. 39th IEEE PVSC,

Tampa, FL, USA, (2013), pp. 0706-0711.

[8] H. Stiebig, C. Zahren and U. Rau, Simple Approach to Characterise Thin-film Silicon Tandem Cell Modules,

Proc. 25th EU PVSEC / 5th WCPEC, Valencia, Spain, (2010), pp. 2744-2747 .

[9] IEC 60904-3 ed 2.0, Photovoltaic devices - Part 3: Measurement principles for terrestrial photovoltaic (PV) solar

devices with reference spectral irradiance data, (2008).

[10] IEC 60904-9 ed 2.0, Photovoltaic devices - Part 9: Solar simulator performance requirements, (2007).

[11] W. Herrmann, Uncertainty of solar simulator spectral irradiance data and problems with spectral match

classification, Proc. 27th

EU PVSEC, Frankfurt, Germany, (2012), pp. 3015 – 3021.

[12] ASTM E 927-10, Standard Specification for Solar Simulation for Terrestrial Photovoltaic Testing, (2010).

[13] JIS C 8912:1998/AMENDMENT 1:2005, Solar simulators for crystalline solar cells and modules

(Amendment 1), (2005).

[14] J. Meier, R. Adelhelm, M. Apolloni, K. Keller, Z. Seghrouchni, D. Romang, S. Benagli, U. Kroll, Traceable

flasher-based power measurement of MicromorphTM

tandem modules, Proc. 26th

EU PVSEC, Hamburg, Germany,

(2011), pp. 2744-2749.

[15] A. Virtuani, H. Müllejans and E. D. Dunlop, Comparison of indoor and outdoor performance measurements of

recent commercially available solar modules, Prog. Photovolt: Res. Appl. 19, (2011), pp. 11-20.

[16] J. Meier, M. Apolloni, K. Keller, R. Adelhelm, D. Romang, S. Benagli, I. Sinicco, Power measurement analysis

of MicromorphTM

tandem modules, Proc. 28th

EU PVSEC, Paris, (2013), pp. 3513 – 3518.

[17] A. Virtuani, D. Pavanello and G. Friesen, Overview of temperature coefficients of different thin film photovoltaic

technologies, Proc. 25th EU PVSEC / 5th WCPEC, Valencia, (2010), pp. 4248 – 4252.

[18] A. Virtuani, M. Pravettoni and H. Müllejans, Comparison of indoor and outdoor performance measurements on

micromorph (a-Si/µc-Si) thin-film solar modules, Proc. 23rd EU PVSEC, Valencia, Spain, (2008), pp. 2378-2382.

[19] IEC 60904-8 ed 2.0, Photovoltaic devices - Part 8: Measurement of spectral response of a photovoltaic (PV)

device, (1998).

[20] J. Burdick and T. Glatfelter, Spectral response and IV measurements of tandem amorphous-silicon alloy solar

cells, Solar Cells 18, (1986), pp. 301-314.

[21] M. Meusel, C. Baur, G. Létay, A. W. Bett, W. Warta and E. Fernandez, Spectral response measurements of

monolithic GaInP/Ga(In)As/Ge triple-junction solar cells: Measurement artifacts and explanation, Prog. Photovolt:

Res. Appl. 11, (2003), pp. 499-514.

[22] M. Pravettoni, To Bias or Not to Bias? An 'How-To' Guide for Spectral Response Measurements of Thin Film

Multi-Junction Photovoltaic Modules, MRS Proceedings 1426, (2012), pp. 81-86.

[23] M. A. Steiner and J. F. Geiz Non-linear luminescent coupling in series-connected multijunction solar cells Appl.

Phys. Lett. 100, 251106 (2012), pp. 1-5.

[24] M. A. Steiner, S. R Kurtz, J. F. Geisz, W. E. McMahon and J. M. Olson Using phase effects to understand

measurements of the quantum efficiency and related luminescent coupling in a multijunction solar cell, IEEE

Journal of Photovoltaics 2(4), (2012), pp. 424-433.

[25] M. A. Steiner, J. F. Geisz, T. E. Moriarty, R. M. France, W. E. McMahon, J. M. Olson, S. R. Kurtz and D. J.

Friedman, Measuring IV curves and subcell photocurrents in the presence of luminescent coupling, Journal of

Photovoltaics 3, (2013), pp. 879-887.

[26] Y. Hishikawa, Y. Tsuno and K. Kurokawa, Spectral response measurements of PV modules and multi-junction

devices, Proc. 22nd EU PVSEC, Milan, (2007), pp. 2765-2769.

[27] M. Pravettoni, A. Komlan, R. Galleano, H. Müllejans and E. D. Dunlop, An alternative method for spectral

response measurements of large‐area thin‐film photovoltaic modules, Prog. Photovolt: Res. Appl. 20, (2012),

pp. 416–422.

31

[28] M. Pravettoni, M. Nicola and G. Friesen, A Filtered Pulsed Solar Simulator for Spectral Response

Measurements of Multi-junction Modules of Commercial Size, Proc. 27th EU PVSEC, Frankfurt, (2012), pp. 3209-

3213.

[29] IEC 60904-7 ed 3.0, Photovoltaic devices - Part 7: Computation of the spectral mismatch correction for

measurements of photovoltaic devices, (2008).

[30] ASTM E2236-10, Standard Test Method for Measurement of Electrical Performance and Spectral Response of

Nonconcentrator Multijunction Photovoltaic Cells and Modules, (2010).

[31] JIS C 8943:2009, Indoor measuring method of output power for multijunction solar cells and modules

(Component reference cell method), (2009).

[32] ECSS-E-ST-20-08C, European Cooperation for Space Standardization, Space engineering, Photovoltaic

assemblies and components, ESA-ESTEC, Nordwijk, Netherland, (31 July 2008).

[33] M. Meusel, R. Adelhelm, F. Dimroth, A. W. Bett and W. Warta, Spectral mismatch correction and spectrometric

characterization of monolithic III-V multi-junction solar cells, Prog. Photovolt: Res. Appl. 10(4), (2002), pp. 243-255.

[34] F. Baumgartner, Power rating multi-junction solar cells: Focus Thin Film; Thin Film Industry Forum 2010, Berlin,

(April 22nd 2010).

32

4 Field performance

4.1 Introduction

Thin-film PV modules display a wide variety of behaviours outdoors because of their varied

spectral, irradiance and temperature responses and their metastable nature. Predicting the

performance of a module based on the conditions outdoors requires an understanding of

these effects. This insight can be achieved through a first-principles approach, through

empirical models derived solely from measured data, or from a combined strategy.

The following contribution summarises literature concerning the spectral, temperature and

transient responses of thin-film PV modules. It is followed by results from a new method of

collecting and analyzing data from diverse geographic locations and from multiple PV

technologies. Next is a treatment of spectral performance analysis using the average photon

energy method on measured spectra, followed by analysis using the average photon

wavelength method on simulated spectra.

4.2 Summary of previous studies in the literature

A Louwen and WGJHM van Sark, Utrecht University, Copernicus Institute of Sustainable Development,

the Netherlands

4.2.1 Introduction

Standard testing procedures for solar panels prescribe that the power output of panels is rated

at standard testing conditions (STC; an overview of standards is given in Ossenbrink et al. [1]).

These standard conditions refer to a module temperature of 25°C, 1000 W/m2 irradiance and

an AM1.5 solar spectrum. Outdoor conditions rarely correspond to these standard values.

Therefore, many studies have sought to establish the performance of PV modules under

realistic operating conditions, by performing outdoor measurements over longer periods of

time.

4.2.2 Spectral effects

Solar cells are rated under spectral conditions according to an air mass of 1.5. However, due

to seasonal and intra-day variations of the solar altitude above the horizon, air mass (AM), and

thus incident spectrum, show considerable variations. The effect of spectral variations on the

performance of crystalline silicon devices is reasonably limited. However, the spectral

response of thin-film devices is different from that of crystalline silicon devices, especially for

amorphous silicon. Fig. 19 shows the spectral response of thin-film solar cells, compared to

several types of crystalline silicon solar cells, as well as the spectrum emitted by a solar

simulator (indicated with ‘Lapss’). The effect of spectral variations on device performance is

measured in different ways in the literature reviewed here. Performance is often expressed in

terms of short circuit current Isc or the field-output-factor (FOF), a parameter relating the actual

performance to the performance one would expect considering STC efficiency and the

irradiance intensity, with corrections for temperature. However, some authors assess

performance by showing the useful fraction (UF) of light with varying spectrum. The useful

fraction is defined in Ref. [2] as “the ratio of the observed spectral irradiance in the useful

33

spectral range of the PV device to the observed global irradiance.” Spectral variations are

otherwise expressed in either air mass units or average photon energy (APE). Data from a

testing facility in Utrecht (UPOT, see Van Sark et al. [3]) has shown APE to decrease with

increasing AM unit.

Fig. 19: Spectral response of different module types (CIS, a-Si, c-Si) [4].

Out of the studied thin-film photovoltaic (TFPV) technologies, the effect of spectral variations

on performance is most pronounced for amorphous silicon. The approach of researchers in

establishing this dependence varies slightly. Some authors have found the performance of a-

Si devices, measured in Isc, to decrease with increasing air mass [4–6], and state that this

effect is caused by the narrow spectral response of a-Si [4], [5], [7]. This trend is shown in Fig.

20. Fanni et al. [5] furthermore show that the effect of spectral variation is larger for triple-

junction a-Si devices compared to single-junction devices, due to current mismatch (see

Krishnan et al. [8]). Other authors analyse the spectral dependence of performance based on

the FOF and APE and find the FOF to increase with increasing APE [9–11]. Minemoto et al.

[10] furthermore establish that APE is indeed a good parameter to analyse the variation of

spectrum with a single characteristic measure. Most of the authors mentioned imply a linear

response of Isc or FOF to spectral variation (APE or AM). However, we could expect a similar

trend as that we see for c-Si: increased performance with increasing APE to a certain

maximum performance, after which further increases of APE lead to a decrease in

performance. Research correlating FOF to APE shows maximum performance for an APE of

about 2 eV, which coincides nicely with the peak in the spectral response of a-Si at about 600

nm. Furthermore the data in Minemoto et al. [10] seem to indicate a decreasing trend after the

2 eV point. However, under normal conditions APEs higher than 2 eV are probably not often

measured under higher irradiance conditions that are needed to perform the measurements

(in order to exclude low irradiance effects).

The spectral response of CdTe solar cells is less limited compared to a-Si PV, but still

somewhat more narrow compared to c-Si devices. Gottschalg et al. [2] have shown the useful

34

fraction for CdTe devices to vary from +4% to –6% around the annual average. It was also

shown that the efficiency of PV modules increases with increasing useful fraction [12].

For CIS/CIGS devices the spectral response is even broader compared to conventional c-Si

devices. As expected, this broad response results in spectral variations having the least effect

on device performance of the TFPV technologies addressed here. Gottschalg et al. show that

for CIS, the useful fraction of the incident light varies by only 5% to 10% during a typical day

[2]. Others mention that the effect of spectral variations on CIS performance is small, but

quantification of performance change is lacking [4], [13]. Kenny et al. [4] state that the Isc of

CIS devices slightly increases with increasing air mass.

Fig. 20: Modelled dependence of Isc on Air Mass of incident radiation, for single junction a-Si

(blue circles) and crystalline silicon (red squares). The dark lines are linear fits to the modeled

data points [6].

4.2.3 Temperature effects

The standard test conditions for PV panels refer to a module temperature of 25°C, 1000 W/m2

irradiance and an AM1.5 solar spectrum. As we can see from the normal operating cell

temperature (NOCT) specified by many PV suppliers, higher module temperatures are easily

reached. For most panels, irrespective of technology, the operating temperature of cells at

800 W/m2 irradiance, 20°C ambient temperature and a wind speed of 1 ms is about 45°-50°C.

For basically all PV technologies, fundamental material physics implies performance loss with

increasing operating temperatures. This notion is confirmed by manufacturers with laboratory

testing designed specifically to establish temperature coefficients for important operating

parameters such as Voc, Vmp, Isc, Imp, and, combining those, Pmax (and thus efficiency).

However, to determine if this temperature dependence of performance holds under realistic

operating conditions, much research has been performed to test these laboratory-established

values.

Makrides et al [14] state that operation at non-STC temperatures results in an average

performance loss of 5% for thin-film technologies. Also, it was found that temperature

coefficients measured outdoors are generally in accordance with supplier specified data [15].

For amorphous silicon devices, temperature-related performance losses are significant, but

much smaller compared to crystalline silicon cells. Of all the thin-film technologies, a-Si is

35

probably the most studied. Various researchers have found the temperature coefficient of

power output to be in the order of –0.2%/K [5], [15–17]. However, as will be discussed below,

because of thermal annealing, higher temperatures can also have positive effects on a-Si

performance. CdTe solar cells are generally found to have a slightly larger temperature

coefficient, but the effect is still considerably smaller compared to crystalline silicon solar cells

at about –0.25%/K [15], [18]. Temperature coefficients for CIGS devices approach the values

commonly observed for c-Si modules. Typically, the temperature coefficient is about –0.36%/K

to –0.42%/K [15].

4.2.4 Transient and seasonal performance changes

Seasonal variations in TFPV performance result from an interaction of different performance

parameters that occur within the same period, but in different phases compared to each other.

As is illustrated in Fig. 21, reflection- and spectrum- related losses have a similar timing, due

to the relation between angle-of-incidence and air-mass. Temperature effects occur inversely

compared to the spectral effects, because it is a general phenomenon that increased

temperature leads to decreased performance (excluding the effect of thermal annealing).

Various papers were reviewed that specifically assessed the seasonal performance changes

of TFPV technologies.

Amorphous silicon

For amorphous silicon modules, seasonal variations are mainly caused by two parameters:

temperature and spectrum. Temperature variations lead to changes in performance through

two effects on the a-Si devices: decrease of performance due to the negative temperature

coefficient and increase of performance due to thermal annealing. As is also shown in Fig. 21,

temperature effects would cause a decrease in performance in warmer periods, and an

increase in colder periods. Also shown in this figure is the inverse but slightly out-of-phase

effect of annealing. Warm temperatures lead to a partial restoration of the degradation that

occurs in a-Si devices. Through separation of the effect of different environmental drivers,

researchers have shown increasing temperatures to lead to performance decreases in

summer because of increased recombination, while performance increases due to thermal

annealing effects [5, 6, 14, 19].

Spectral variations lead to performance peaks in summer, because spectra are more

favorable (blue-shift) in summer months compared to winter months (red-shift). See also Fig.

20. For a-Si, spectral effects are especially pronounced and various research has confirmed

this seasonal spectral effect [5, 12, 15]

Other seasonal effects that are mentioned is the decreased reflectance caused by the lower

solar angle in summer [5], although the study by Fanni et al. [5] analysed a horizontally

orientated PV installation, which would logically result in increased reflectance in winter

compared to tilted installations.

The resulting total seasonal performance is the result of the interplay of these factors. The

strength of each seasonal driver on overall seasonal variation seems to be somewhat related

to the installed location. Gottschalg et al. [19] and Virtuani et al. [6] state that spectral

variations are the main reason for the performance peak observed in summer. Other

researchers do find evidence of the influence of spectral changes on seasonal performance,

but consider the positive performance effect of thermal annealing the most important driver of

36

summertime performance peaks [14]. The result is similar however, because the effect of

spectrum and annealing overlap and are only slightly out of phase, resulting in performance

maxima in summer and minima in winter. This trend is shown in Fig. 21.

Fig. 21: Left: Overview of the seasonal effects of temperature (εt), SWE (εswe), reflection (εr)

and spectrum (εs) on a-Si performance. The effects are expressed as performance factors, the

ratio of actual vs. rated performance due to each factor. The total performance factor is obtained

by multiplication of the four individual performance factors [5]. Right: Seasonal changes of

selected amorphous silicon devices [19].

CdTe and CIGS

For CdTe and CIGS modules, the seasonal performance changes are less pronounced

compared to a-Si technology. Because of the limited effect of spectral variation, CIGS

modules exhibit seasonal behaviour very comparable to that of c-Si modules. The main factor

for CIGS devices is the decrease in summer performance caused by the relatively high

dependence of performance on operating temperature [14]. For CdTe, a similar seasonal

variation is observed, although the variations are less pronounced because of the lower

temperature coefficient [14]. Seasonal performance (monthly DC performance ratio) of CIGS

and CdTe modules is shown in Fig. 22.

Fig. 22: Monthly average DC performance ratio for thin-film CIGS and CdTe systems over the

period June 2006 – June 2010 in Nicosia, Cyprus [14].

37

References for this section

[1] H. Ossenbrink, H. Müllejans, R. Kenny, and E. Dunlop, 1.39 - Standards in Photovoltaic Technology, in

W.G.J.H.M. van Sark (ed), Photovoltaic Technology Volume 1A in Comprehensive Renewable Energy, A. Sayigh,

Ed. Oxford: Elsevier, (2012), pp. 787–803. [2] R. Gottschalg, D. G. Infield, and M. J. Kearney, Experimental study of variations of the solar spectrum of

relevance to thin-film solar cells, Solar Energy Materials and Solar Cells, vol. 79, no. 4, (Sep. 2003), pp. 527–537. [3] W. G. J. H. M. van Sark, A. Louwen, A. C. de Waal, B. Elsinga, and R. E. I. . Schropp, UPOT: The Utrecht

Photovoltaic Outdoor Test Facility, Proc. 27th EU PVSEC, (2012), pp. 3247–3249. [4] R.P. Kenny, A. Ioannides, H. Müllejans, W. Zaaiman, E.D. Dunlop, Performance of thin film PV modules, Thin

Solid Films, Volumes 511–512, (26 July 2006), pp. 663-672. [5] L. Fanni, A. Virtuani, and D. Chianese, A detailed analysis of gains and losses of a fully-integrated flat roof

amorphous silicon photovoltaic plant, Solar Energy, vol. 85, no. 9, (Sep. 2011), pp. 2360–2373. [6] A. Virtuani and L. Fanni, Seasonal power fluctuations of amorphous silicon thin-film solar modules:

distinguishing between different contributions, Progress in Photovoltaics: Research and Appications. doi:

10.1002/pip.2257, (2012), pp. 208-217.

[7] M. Muñoz-García and O. Marin, Characterization of thin-film PV modules under standard test conditions:

Results of indoor and outdoor measurements and the effects of sunlight exposure, Solar Energy, vol. 86, no. 10,

(Oct. 2012), pp. 3049–3056. [8] P. Krishnan, J. W. A. Schüttauf, C. H. M. van der Werf, B. Houshyani Hassanzadeh, W. G. J. H. M. van Sark,

and R. E. I. Schropp, Response to simulated typical daily outdoor irradiation conditions of thin-film silicon-based

triple-band-gap, triple-junction solar cells, Solar Energy Materials and Solar Cells, vol. 93, no. 6–7, (Jun. 2009),

pp. 691–697. [9] S. Nagae, M. Toda, T. Minemoto, H. Takakura, and Y. Hamakawa, Evaluation of the impact of solar spectrum

and temperature variations on output power of silicon-based photovoltaic modules, Solar Energy Materials and

Solar Cells, vol. 90, no. 20, (Dec. 2006), pp. 3568–3575. [10] T. Minemoto, Y. Nakada, H. Takahashi, and H. Takakura, Uniqueness verification of solar spectrum index of

average photon energy for evaluating outdoor performance of photovoltaic modules, Solar Energy, vol. 83, no. 8,

(Aug. 2009), pp. 1294–1299. [11] C. Sirisamphanwong and N. Ketjoy, Impact of spectral irradiance distribution on the outdoor performance of

photovoltaic system under Thai climatic conditions, Renewable Energy, vol. 38, no. 1, (Feb. 2012), pp. 69–74. [12] R. Gottschalg and T. Betts, On the importance of considering the incident spectrum when measuring the

outdoor performance of amorphous silicon photovoltaic devices, Measurement Science and Technology, vol. 15,

no. 2, (Feb. 2004), pp. 460–466. [13] W. Okullo, M. K. Munji, F. J. Vorster, and E. E. van Dyk, Effects of spectral variation on the device

performance of copper indium diselenide and multi-crystalline silicon photovoltaic modules, Solar Energy

Materials and Solar Cells, vol. 95, no. 2, (Feb. 2011), pp. 759–764. [14] G. Makrides, B. Zinsser, A. Phinikarides, M. Schubert, and G. E. Georghiou, Temperature and thermal

annealing effects on different photovoltaic technologies, Renewable Energy, vol. 43, (Jul. 2012) pp. 407–417. [15] G. Makrides, B. Zinsser, G. E. Georghiou, M. Schubert, and J. H. Werner, Temperature behaviour of different

photovoltaic systems installed in Cyprus and Germany, Solar Energy Materials and Solar Cells, vol. 93, no. 6–7,

(Jun. 2009), pp. 1095–1099. [16] D. Meneses-Rodríguez, P. P. Horley, J. González-Hernández, Y. V. Vorobiev, and P. N. Gorley, Photovoltaic

solar cells performance at elevated temperatures, Solar energy, vol. 78, (Feb. 2005), pp. 243–250. [17] T. Ishii, K. Otani, and T. Takashima, Effects of solar spectrum and module temperature on outdoor

performance of photovoltaic modules in round-robin measurements in Japan, Progress in Photovoltaics: Research

and Applications, vol. 19, no. 2, (Mar. 2011), pp. 141–148. [18] S. Dittmann, W. Durisch, and J. Mayor, Comparison of outdoor and indoor characterization of a CdTe PV

module, Proc. 25th

EU PVSEC, (Sep. 2010), pp. 3508–3512. [19] R. Gottschalg, T. Betts, and S. Williams, A critical appraisal of the factors affecting energy production from

amorphous silicon photovoltaic arrays in a maritime climate, Solar energy, vol. 77, no. 6, (Dec. 2004), pp. 909–

916. [20] M. Gostein and L. Dunn, Light soaking effects on photovoltaic modules: Overview and literature review, Proc.

37th IEEE PVSC, Seattle, WA, USA, (2011), pp. 3126–3131. [21] K. Astawa, T. R. Betts, and R. Gottschalg, Effect of loading on long term performance of single junction

amorphous silicon modules, Solar Energy Materials and Solar Cells, vol. 95, (2011), pp. 119–122.

38

4.3 Comparison of Different PV Module Technologies: New Method

to Analyse Performance Characteristics Obtained from Field

Tests at Different Locations

Ulrike Jahn, TÜV Rheinland, Germany Gabi Friesen, University of Applied Sciences of Southern Switzerland (SUPSI)

4.3.1 Motivation

The energy yield measurements of PV modules at different climatic locations play an

important role for the economic operation of PV systems. Assessment of the PV module

performance must be investigated under real operating conditions rather than based on data

sheet information given by the manufacturer. In recent years, research and industry have

developed outdoor test facilities to better understand the climate-dependent performance of

PV modules and to gain comparative results of energy yield testing of different technologies.

In order to make use of the wealth of available monitoring data in various locations collected

by different labs worldwide [1-4], we have undertaken a joint effort to develop a common

method to analyse PV module field data obtained from different test facilities and locations.

4.3.2 Approach

Outdoor PV module measurements from seven different locations in Germany, Switzerland,

France, Norway, UK, Colorado/USA and Cyprus, are analyzed and compared to demonstrate

the new approach. Fig. 23 shows the irradiation profiles and average daylight ambient

temperatures of each site. The exposed PV modules from different manufacturers and

production years include various thin-film technologies (a-Si, a-Si/μ-Si, a-Si/a-Si, a-Si/a-Si/a-Si,

CIS, CIGS, CdTe) and crystalline Si modules (mono and poly).

The seven outdoor test facilities use different measurement setups (based on MPP tracking or

I-V curve tracing), frequency of data recording (10 seconds, 5 minutes, 10 minutes,

15 minutes), cleaning procedures of sensors and modules as well as different periods of

outdoor measurements. For this data analysis, the same meteorological parameters are used

for the comparison: in-plane irradiance measured by pyranometer Gpyr, average ambient

temperature Tamb and average back-of-module temperature Tbom. The Pmax,STC and Isc,STC

values of the PV modules are determined by outdoor or indoor measurements or the

nameplate rating is used.

In order to compare the performance characteristics of the modules at the different sites, a

common data format was defined (Tab. 2). In a first step, daily performance ratios of Pmax and

of Isc are used as key parameters to characterise the module performance. Spectral data, if

available, will be included in a second stage.

More outdoor performance data from labs worldwide is being collected and analysed, while

the methods for data evaluation and the result plots for the comparison of different PV module

technologies will be further improved.

39

Fig. 23: Meteorological data during monitoring period 2010-2012 at seven different sites. The

upper figure shows the different distributions of daily irradiation sums, the lower figure shows

the monthly average ambient temperatures measured at the sites.

Tab. 2: Parameters defined and used to perform common data analysis

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

1000 2000 3000 4000 5000 6000 7000 8000 9000

Ene

rge

tic

dis

trib

uti

on

of d

aily

irra

dia

tio

n

Daily irradiation [Wh/m²]

Norway

UK

Germany

USA

Switzerland

France

Cyprus

-5

0

5

10

15

20

25

30

35

40

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

ave

rage

dai

ly a

mb

ien

t te

mp

era

ture

Norway

UK

Germany

USA

Switzerland

France

Cyprus

40

4.3.3 Results

The performance ratio PR(Pmax) is the most general performance parameter to characterise

PV module and system performance [5]. Fig. 24 shows an example of data representation for

some crystalline silicon modules monitored at five different sites.

Fig. 24: Average back-of-module temperature for irradiances G > 0 W/m² (line plots) and daily PR

for Pmax (dot plots) for c-Si modules at five different sites during monitoring period May 2010 to

Dec 2012

Initially, the PR(Pmax) and PR(Isc) performance ratio as defined in Tab. 1 are plotted into 1-year

plots (if no degradation has been detected over the observed period). Fig. 25 shows the step-

by-step approach of how data are analysed. The plots represent different crystalline silicon

modules.

Whereas the top/right plot in Fig. 25 shows the typical seasonal variations of PR(Pmax) caused

by module temperature, the left plot representing PR(Isc) is almost stable because of the very

low spectral dependency of c-Si [2]. The PR(Isc) variations are in the range of ±5%, and they

are mainly the sum of the uncertainties in irradiance measurement (pyranometer calibration),

determination of Isc at STC and very small spectral variations for c-Si modules. To be able to

exclude these uncertainties or influences from further analysis, the first step consists of a

correction in which PR(Isc) data are set equal to one, followed by application of the so-obtained

correction factor to the PR(Pmax) performance ratio. The result of this normalization procedure

is shown in plot 3 (middle graph), which then allows analysis of the low-irradiance behaviour in

more detail without spectral effects and the pure temperature behaviour as shown in the two

lower plots in Fig. 25.

41

Fig. 25: Example of analysis approach with daily performance ratio PR(Pmax) (top/right) and

PR(Isc) (top/left) of 6 different c-Si modules measured at different locations. Step 1: merge to

normalised performance ratio norm. PR(Pmax)and Step 2: plot in dependence of daily irradiation

(bottom/left) and of average daylight module temperature (bottom/right).

0.2

0.4

0.6

0.8

1

1.2

Jan Feb Apr May Jul Sep Oct Dec

(ΣIs

c/Is

c,st

c)/(

ΣG/1

00

0)

Norway

UK

Germany

USA

Switzerland

France

0.2

0.4

0.6

0.8

1

1.2

Jan Feb Apr May Jul Sep Oct Dec

No

rm.

(ΣP

m/P

m,s

tc)/

(ΣG

/10

00

)

Norway

UK

Germany

USA

Switzerland

France

0.2

0.4

0.6

0.8

1

1.2

0 1'000 2'000 3'000 4'000 5'000 6'000 7'000 8'000 9'000

No

rm.

(ΣP

m/P

m,s

tc)/

(ΣG

/10

00

)

Daily irradiation H [Wh/m²]

NorwayUKGermanyUSASwitzerlandFrance

0.2

0.4

0.6

0.8

1

1.2

0 10 20 30 40 50

No

rm.

(ΣP

m/P

m,s

tc)/

(ΣG

/10

00

)

daylight avg. back of module temperature [°C]

NorwayUKGermanyUSASwitzerlandFrance

STEP1

0.2

0.4

0.6

0.8

1

1.2

Jan Feb Apr May Jul Sep Oct Dec

(ΣP

m/P

m,s

tc)/

(ΣG

/10

00

)

NorwayUKGermanyUSASwitzerlandFrance

STEP2

42

Fig. 26 shows different amorphous silicon module data (a-Si, a-Si/μ-Si, a-Si/a-Si, a-Si/a-Si/a-

Si) monitored at the seven different sites.

It is well known that amorphous silicon (a-Si) based modules are subject to seasonal

variations of performance between 0% to 15% around an average value, which depends on

the module technology, local climatic conditions and type of integration [6]. The seasonal

variations result from two counteracting effects — light-induced degradation and thermal-

induced recovery — also known as the Staebler Wronski effect [7]. The offsets in performance

ratio shown in Fig. 26 are caused by the difficulty of determining a unique reference STC

power for amorphous silicon. Today no standard approach exists on how to determine a

reproducible and unique reference value.

Fig. 26: Average back-of-module temperature for irradiances G > 0 W/m² (line plots) and daily

PR of Pmax (dot plots) for a-Si based modules measured in different locations during monitoring

period May 2010 to Dec 2012.

Fig. 27 shows a similar data representation to the c-Si data presented before, now for a

selection of 7 different amorphous silicon based modules (1 per location). The first two plots at

the top of the figure show how the performance ratio PR(Pmax) and PR(Isc) display seasonal

variations due to the spectrum, with increasing PR values in summer and decreasing values in

winter months. In contrast to c-Si, the Si PV modules display an opposite PR(Pmax) seasonal

variation and also show a seasonal variation of the PR(Isc) value. The spectrum has a

significant impact on both the PR(Pmax) and PR(Isc) plots. The normalization step leads to a

correction of the spectral effects, and thermal effects are therefore highlighted. The two

bottom graphs obtained after the normalization represent the typical low-irradiance and

temperature behaviour of amorphous silicon modules.

-10

10

30

50

70

90

110

0.1

0.3

0.5

0.7

0.9

1.1

1.3

Apr 10 Jul 10 Oct 10 Jan 11 May 11 Aug 11 Nov 11 Mar 12 Jun 12 Sep 12 Dec 12

Ave

rage

bac

k-o

f-m

od

ule

tem

per

atu

re [

°C]

(ΣP

m/P

m,s

tc)/

(ΣG

/100

0)

NREL(1j) SUPSI(1j) CREST(1j) EURAC(1j) NREL(2j) SUPSI(2j) UCY(2j)

INES(3j) UiA(3j) CREST(3j) TUV(µc-Si) SUPSI(µc-Si) NREL(other)

43

Fig. 27: Example of analysis approach with daily performance ratio PR(Pmax) (top/right) and

PR(Isc) (top/left) of 7 different a-Si based silicon modules measured in different locations. Step 1:

merge to normalised performance ratio norm PR(Pmax) and Step 2: plot in dependence of daily

irradiation (bottom/left) and of average daylight module temperature (bottom/right).

0.2

0.4

0.6

0.8

1.0

1.2

Jan Feb Apr May Jul Sep Oct Dec

(ΣIs

c/Is

c,st

c)/(

ΣG/1

00

0)

NorwayUKGermanyUSASwitzerlandFranceCyprus

0.2

0.4

0.6

0.8

1.0

1.2

Jan Feb Apr May Jul Sep Oct Dec

no

rm. (

ΣPm

ax/P

max

,stc

)/(Σ

G/1

00

0)

Norway

UK

Germany

USA

Switzerland

France

Cyprus

0.2

0.4

0.6

0.8

1.0

1.2

0 1'000 2'000 3'000 4'000 5'000 6'000 7'000 8'000

No

rm. (

ΣPm

ax/P

max

,stc

)/(Σ

G/1

00

0)

Daily irradiation H [Wh/m²]

NorwayUKGermanyUSASwitzerlandFranceCyprus

0.2

0.4

0.6

0.8

1.0

1.2

Jan Feb Apr May Jul Sep Oct Dec

(ΣP

max

/Pm

ax,s

tc)/

(ΣG

/10

00

)

Norway

UK

Germany

USA

Switzerland

France

Cyprus

0.2

0.4

0.6

0.8

1.0

1.2

0 10 20 30 40 50

No

rm. Σ

Pm

ax/

Pm

ax,

stc)

/(ΣG

/10

00

)

daylight avg. back of module temperature [°C]

NorwayUKGermanyUSASwitzerlandFranceCyprus

44

In contrast to c-Si modules, a-Si PV modules show an opposite PR(Pmax) seasonal variation

and a seasonal variation of the PR(Isc) value. The spectrum has a significant impact on both

plots, the PR(Pmax) and PR(Isc). The normalization step leads to a correction of the spectral

effects, and thermal effects are thus highlighted. The two bottom graphs obtained after the

normalization represent the typical low-irradiance and temperature behaviour of amorphous

silicon modules.

References for this section

[1] G. Friesen. et al., Detailed Analysis of thin-film Module Performance Based on Indoor and Outdoor Tests, Proc.

27th

EU PVSEC, Frankfurt, Germany, (2012), pp. 3027 - 3032.

[2] U. Jahn et al., Final Results of High Precision Indoor and Outdoor Performance Characterization of Various

Thin-Film PV Module Technology, Proc. 27th EU PVSEC, Frankfurt, Germany, (2012), pp. 3233–3238.

[3] J. Merten, et al., Diagnostics of Degradation of thin-films PV Modules During 3.85 Years of Outdoor Exposure,

Proc. 27th EU PVSEC, Frankfurt, Germany, (2012), pp. 3219 - 3224.

[4] Imenes A.G., Yordanov, G.H., Midtgård. O.M., Saetre, T.O. Development of a test station for accurate in situ I-V

curve measurements of photovoltaic modules in Southern Norway, 37th IEEE PVSC, Seattle, WA, USA, (2011), pp.

003153-003158.

[5] IEC 61724, Ed 1.0. Photovoltaic system performance monitoring - Guidelines for measurement, data exchange

and analysis. (November 1998).

[6] L. Fanni at el., A detailed analysis of gains and losses of a fully-integrated flat roof amorphous silicon

photovoltaic plant. Solar Energy (2011); 85: pp. 2360–2373.

[7] Various Authors. Thin-film Silicon Solar Cells. Shah A (ed.), EPFL Press, (2010).

4.4 Analysis of spectral effects on outdoor performance

Markus Schweiger, TÜV Rheinland

4.4.1 Spectral response measurements

The spectral behaviour of PV modules can be described by the external quantum efficiency

(EQE). It quantifies the average number of photons per wavelength Φ(λ) which are required

for generating one electron by lifting the electron from the valence to the conduction band. The

number of electrons is calculated by dividing the current I by the elementary charge q. Taking

into account the wavelength dependence of the energy of the different photons, the spectral

response (SR) of the device in AW–1nm-1 can be calculated from the external quantum

efficiency. The energy is calculated in terms of the Planck constant h and the frequency f or,

inversely, the wavelength λ.

(1.1)

Measurements of the spectral dependencies of the short circuit current Isc differ in measuring

with constant photon flux or at constant irradiance. Both methods describe the wavelength

intervals in which the cells perform best and are able to generate photocurrent for electrical

energy. Generalizing, we can say that the performance in the short-wavelength range

depends on the top layers of the module, such as the anti-reflection coating, the glass, the

encapsulation material, TCO, the SiOX layers and upper cell structures. In contrast, the

performance at longer wavelengths depends on the lower cell regions and the module

temperature. The measurement of the spectral response is important to validate the observed

45

outdoor spectral behaviour by integrating the solar spectral irradiance with the spectral

response.

(1.2)

Spectral response measurements must be performed according to IEC 60904-8 [3].

Differences between the investigated technologies (CdTe, CIGS, a-Si, a-Si/µc-Si, c-Si) as

within one technology (a-Si/µc-Si, CIGS) have been observed. Representative results are

shown in Fig. 28.

Fig. 28: Relative spectral response signal of different modules, illustrating significant

differences within the same thin-film technology

4.4.2 Measuring solar spectral irradiance

For spectral analysis, real solar spectral irradiance data are essential. The spectral conditions

in the module plane must be known. These conditions depend on the mounting direction and

the location of the test site. Moreover, high data quality is needed. The spectral distribution

must be measured fast (<1 s) to reduce the influence of irradiance fluctuations and at least in

an interval of (300–1300) nm to be relevant to all technologies, some of which have SR-

signals up to 1300 nm (CIGS). Therefore, a combination of a Si-CCD-sensor and an InGaAs-

CCD-sensor is needed. Regular calibration of the measurement system at least once a year

and a daily cleaning of the input-optics are essential. The input optic and its angle dependency

have been identified as one of the crucial factors for spectral measurements, as shown in

Fig. 29 and 30.

46

Fig. 29: Angular characteristic of the best performing input optic: spectrally neutral but

increasing cosine error above 40°

Fig. 30: Angular characteristic of the worst performing input optic: wavelength dependent

cosine error

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

-90 -80 -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90Co

sin

e c

orr

ecte

d a

nd n

orm

alis

ed s

ignal

Angle of incidence [°]

400-500 nm

500-600 nm

600-700 nm

700-800 nm

800-900 nm

900-1000 nm

1000-1100 nm

1100-1200 nm

1200-1300 nm

1300-1400 nm

1400-1500 nm

1500-1600 nm

Cosine

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

-90 -80 -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90Co

sin

e c

orr

ecte

d a

nd

no

rma

lise

d s

ign

al

Angle of incidence [°]

400-500 nm

500-600 nm

600-700 nm

700-800 nm

800-900 nm

900-1000 nm

1000-1100 nm

1100-1200 nm

1200-1300 nm

1300-1400 nm

1400-1500 nm

1500-1600 nm

cosine

47

4.4.3 Analysis of large amounts of spectral data with the APE factor

The distribution of the solar spectrum was measured every minute with 1243 data points in the

range of (300–1600) nm. The integration time for collecting a complete spectrum was set to

about two seconds so as to render fluctuations negligible. During the monitoring period

approximately 400,000 spectra were recorded, ruling out manual processing. The raw data

spectrum includes the variation in the absolute value of the irradiance, with a maximum at

mid-day under blue-sky conditions. The data do not yield qualified information about the

blueness or redness of the spectrum suitable for automated data processing. Each spectrum

can be classified according to IEC 60904-9 to determine the rough agreement with STC [2]

(AM1.5) spectral conditions. This is not an adequate way to analyse shifts in the distribution of

the solar spectrum, and the classification is limited to an unsatisfying wavelength interval

between 400 and 1100 nm. To evaluate the data from months or years and to describe the

blueness or redness of the spectra, the calculation of average photon energy ([APE]=eV)

offers an adequate method for simplifying the data and describing a complete spectrum with

just a single value [4]. For an integration range of (300–1600) nm, the reference AM1.5

spectrum has an APE value of 1.65 eV. Spectra with an APE < 1.65 eV can be classified as

red spectra and spectra with an APE > 1.65 eV as blue.

(3.1)

where Ei (λ) is the spectral irradiance [W·m-2·nm-1], the electronic charge qe' is the conversion

factor for obtaining electron-volts instead of joules and Φi (λ) is the spectral photon flux density

[m-2·s-1].

The shapes of the clear-sky APE curves differ for different seasons, as shown in Fig. 31. A

strong blue shift in the morning (before approximately 9 am) and evening (after approximately

6 pm) for summer days was observed, due to the sun rising and setting behind the module

plane. For winter days the sun rises and sets in front of the module plane. Direct sunlight

prevails for nearly the entire winter day and the module must handle large air mass changes

over the course of the day. Declining air mass until midday results in a blue shift of the

spectrum, reaching a maximum at approximately 12:30 pm. Spectral conditions compliant to

the standard can be attained only in spring and fall for modules with fixed mounting.

Increasing vapour and aerosol content in the evening was detected only for days with a high

air mass. Other mounting directions than south yielded an APE curve shaped similarly to a

tangent curve. For cloudy days a strong blue shift in the spectral distribution may be

generalised because of the much higher diffuse light fraction. Partly cloudy days show a

correlation between the appearance of clouds (low irradiance) and the APE values, as shown

in Fig. 32. In characterizing the seasonal conditions, we may generalise and say that the

winter months are redder than the summer months, as shown in Fig. 33. This is also the case

for diffuse days. The rare clear sky days in winter show a strong red shift. Even in winter

diffuse days are very frequent, summer is bluer than winter.

48

Fig. 31: Solar spectral irradiance drift for four cloudless days in summer (red), spring (orange),

all (yellow) and winter (blue), mounted facing south and tilted 35°, in Cologne

Fig. 32: Characterization of a partly cloudy day (29.08.2010) and correlation of APE values with

the appearance of clouds

08:00 10:00 12:00 14:00 16:00 18:00 20:00

0,0

1,2

1,3

1,4

1,5

1,6

1,7

1,8

1,9

2,0

2,1

Ave

rag

e P

ho

ton

En

erg

y A

PE

[e

V]

time t [hh:mm]

APE (AM1.5)

Summer

Winter

Spring

Fall

0

200

400

600

800

1000

1200

1.0

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

1.9

2.0

7:00 8:26 9:52 11:19 12:45 14:12 15:38 17:04 18:31

Irra

dia

nce [W

/m²]

Avera

ge P

hoto

n E

nerg

y A

PE

[eV

]

time t [hh:mm]

APE Irradiance

49

Fig. 33: Seasonal variations of APE values and average APE value per month

4.4.4 Correlating SR data and spectral data with module performance

Shifts in the spectral distribution primarily affect the generated photocurrent. The photocurrent

increases linearly with irradiance (as tested and confirmed in the laboratory for all specimens

between (100 and 1100) W/m²). If there is a change in the spectrum, the useful fraction for the

photo-effect changes. This effect can be offset by a change in the irradiance at the module.

Knowing the SR-signals of each module and having measured the real outdoor spectrum, we

can calculate the resulting photocurrent theoretically and analyse the current as a function of

spectral shifts or rather APE.

To associate this theoretical value of Isc with the measured values of the I-V curves, each

curve measured must be corrected for temperature and irradiance relative to comparable

conditions (STC). Assuming the Isc is stable during the measurement period and the modules

are clean, only the influence of spectral changes remains. The standard IEC 60891 [5]

provides appropriate procedures, while procedure No. 2 was selected for this work. Each data

point of an I-V curve must be corrected individually with a total of five different parameters.

The resulting Isc dependence on spectrum for different module technologies is shown in

Fig. 34.

The data should be filtered in order to obtain sharply defined curves of Isc (APE) at least

through a plausibility check. When measuring with a pyranometer, data with angles of

incidence above 50° must be excluded. Above this angle the good cosine behaviour of the

glass dome exceeds the flat module plane, and angle dependence can affect the results.

Moreover, possible angle dependences of the SR signal will be locked out, along with the

dependence of the SR signal on the irradiance level and temperature. The correction range of

the IEC 60891 method should be kept as small as possible and curves at very low irradiation

should be discarded because of disturbing influences, such as the strong dependence of the

temperature coefficients on the irradiance in some thin-film modules.

50

Fig. 34: Dependence of measured Isc on spectral changes (T, G corrected, normalised to ISC, STC

(=100%) at 1.65 eV)

4.4.5 Influence of spectral effects on the energy yield of single-junction

modules

To characterise the periods of days, months and years, an irradiance-weighted average

spectrum was calculated by summing the absolute values of each spectrum and dividing the

result by the number of spectra summed over. The correct weighting of each spectrum is

obtained by correlating the absolute values with the irradiance at the time of measurement.

Assuming that the output power is nearly linear with the short-circuit current, we can quantify

the gains or losses in the energy yield. Further weighting with the nonlinear low-irradiance

characteristics of Pmax has not yet occurred at this point.

The annual average spectrum for the analysed test site is shown in Fig. 35. The spectrum

nearly fits with the IEC 60904-3 [6] spectrum. The spectral influence for standard crystalline

modules is below 1%. Because of the strong dependency of a-Si modules on spectral effects

as shown in Fig. 34, gains of up to 3% are possible. Losses caused by spectral effects for the

tandem module, because of current mismatch, have to be considered separately.

Fig. 35: Average spectrum of the period 01.09.11–31.08.12 relative to AM1.5, area normalised

300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600

Spect

ral I

rradia

nce

Wavelength [nm]

IEC 60904-3_Ed.2, 921 W/m², 1.65 eV

AVG_01.09.11-31.08.12, 355 W/m², 1.68 eV

IEC - AVG

51

References for this section

[1] M. Schweiger et al., Spectral Analysis of Various Thin-Film Modules Using High Precision Spectral Response

Data and Solar Spectral Irradiance Data, Proc. 27th EU PVSEC, Frankfurt, Germany, (2012), pp. 3284 – 3290.

[2] W. Herrmann et al., Uncertainty of Solar Simulator Spectral Irradiance Data and Problems with Spectral Match

Classification, Proc. 27th EU PVSEC, Frankfurt, Germany, (2012), pp. 3015 – 3021.

[3] IEC 60904-8 (CDV), Photovoltaic devices - Part 8: Measurement of spectral response of a photovoltaic (PV)

device, (2012).

[4] T. Betts, Investigation of Photovoltaic Device Operation under Varying Spectral Conditions, Loughborough

University, (2004).

[5] IEC 60891: Photovoltaic devices – Procedures for temperature and irradiance corrections to measured I-V

characteristics, (2009-12).

[6] IEC 60904-3: Photovoltaic devices – Measurement principles for terrestrial photovoltaic (PV) solar devices with

reference spectral irradiance data, (2008-03).

4.5 Modeled spectrum method

G. Belluardo, J. Wagner, A. Tetzlaff, P. Ingenhoven, D. Moser (EURAC)

4.5.1 Introduction

Many parameters are used to characterise the quality of spectral distribution: average photon

energy [1], air mass [2], useful fraction [3] and spectral mismatch correction [4]. The

parameters are based on measured or modeled solar spectra, and may or not require the

measurement of the module spectral response.

The purpose of this work is to present the effect of the solar spectrum on the performance of

different PV technologies using a methodology that does not require availability of ground

measurements of spectral irradiance. The spectral distribution is represented by a unique

value, the average wavelength, which is calculated from modeled spectra using satellite-

retrieved cloud information. The integral value of spectral distribution is validated through

comparison with measured irradiance.

4.5.2 Description of methodology

Modeling of solar spectrum

Spectrally resolved solar surface irradiance is calculated with the SPECMAGIC algorithm [5].

This algorithm produces results with high accuracy and speed because it uses lookup tables

(LUTs), i.e., pre-computed radiative transfer model (RTM) results, which contain the

transmittance for a variety of atmospheric and surface states. In this way, RTM does not need

to be resolved for every satellite pixel and time, but the transmittance can just be extracted

from the LUTs by interpolation. Finally, solar surface irradiance can be calculated from the

transmittance by multiplication with the extraterrestrial incoming solar flux density.

The following input parameters are taken into account:

- aerosol optical depth (from monthly climatologies)

- surface albedo (using land use maps)

- single scattering albedo (fixed value)

- total column ozone (fixed value)

- water vapor column (from monthly climatologies)

- sun-earth distance

52

- solar zenith and azimuth angles

In a first stage, the direct and global clear sky spectral irradiance is calculated with the

SPECMAGIC algorithm. To decrease the computation run-time the correlated-k approach of

Kato et al. is used [6], which divides the whole solar spectrum into 32 bands, each of which is

characterised by a unique set of values of gaseous absorption coefficients. The component j

of

irradiance at generic wavelength λ is derived as follows:

Ij,mod(λ) = Ij,am1.5(λ) * Ij,mod(k) / Ij,am1.5(k)

Where:

- Ij,am1.5(λ): irradiance of the standard spectrum at air mass 1.5, at wavelength λ

- Ij,mod(k): irradiance modeled with SPECMAGIC, at Kato band k

- Ij,am1.5(k): irradiance of the standard spectrum at air mass 1.5, at Kato band k

The direct and global actual spectral irradiance is then obtained by attenuating the clear sky

spectral irradiance according to the actual clouds using the cloud index from MeteoSwiss [7].

The cloud index is converted to the clear sky index and direct and global irradiance are simply

calculated as the product of clear sky irradiance and clear sky index. A height correction is

also performed taking into account the site altitude and solar zenith angle.

The modeled direct and global horizontal spectral irradiance are finally modified in order to

take into account the shadowing of the surrounding mountains, and converted to tilted plane

by applying a model that assumes an isotropic distribution of diffuse radiation and a fixed

value of surface albedo.

Calculation of average wavelength

The average wavelength of a solar spectrum is the wavelength for which the integral value of

irradiance at lower wavelengths equals the integral value of irradiance at higher wavelengths.

In other words, at the average wavelength the cumulative spectral irradiance is exactly half of

the broadband irradiance.

The spectral range considered in this study is between 307 nm and 1965 nm (corresponding

to Kato bands 4 to 27). This is done in order to consider the range of photovoltaic conversion

of state-of-the-art solar cells (about 300 nm to 1700 nm). The average wavelength values of

the AM1.5 standard spectra in the previously mentioned spectral range are 716 nm for direct

irradiance and 699 nm for global irradiance.

Measured weather and PV production data

The photovoltaic technologies considered in this study are installed at the Airport Bolzano

Dolomiti (ABD) test facility in South Tyrol in northern Italy (46.45778 N, 11.32861 E), which

was connected to the medium-voltage grid in August 2010. The plant is composed of a 662

kW commercial section with CdTe modules, and a 62 kW experimental section, with 24

different types of modules, divided into 39 arrays ranging between 1 and 2 kW each, and

mounted on fixed racks as well as on single- and dual axis trackers [8]. The technologies

considered in this study have a fixed tilt of 30° and an azimuth angle of 8.5° West of South.

The site elevation is 262 m above sea level.

53

The production data are recorded by commercial inverters with a frequency of 15 minutes.

The facility is also equipped with a meteo station for the measurement of horizontal (global

and diffuse) and global in-plane-with-modules irradiance (secondary standard pyranometers),

direct irradiance (secondary standard pyrheliometer), and back-of-module temperature (Pt100

sensors).

4.5.3 Results

Validation

The solar irradiance on a horizontal plane and in plane with modules (fixed rack) calculated

with SPECMAGIC for the ABD site, considering the whole spectral range, is validated through

a comparison with the corresponding measured values, for the year 2011. Fig. 36 and Fig. 37

show a good agreement, both for unfiltered points and for points corresponding to clear sky

conditions (ratio of diffuse to global horizontal irradiance lower than 0.20), and for both planes.

Fig. 36: Comparison of horizontal simulated and measured irradiance for ABD test installation,

year 2011. Top: unfiltered points (RMSE=80.2). Bottom: clear sky conditions (RMSE=31.8)

54

Fig. 37: Comparison of in-plane simulated and measured irradiance for ABD test installation,

year 2011. Top: unfiltered points (RMSE=100.870). Bottom: clear sky conditions (RMSE=71.264)

55

Spectral effects

In this section, relations between modeled average wavelength and PV technology

performance are presented for four different thin-film technologies: single-junction amorphous

silicon (a-Si), micromorph silicon (a-Si/μc-Si), Cadmium Telluride (CdTe) and Copper Indium

Gallium Diselenide (CIGS). A polycrystalline silicon technology is also considered as

reference.

The data here presented are filtered:

For Angle Of Incidence (AOI) less than 50°, in order to avoid angle dependence due to

mismatch of cosine behaviour of pyranometer glass dome with respect to flat module

plane

Clear sky conditions

Performance Ratio range given by PR ± σPR to sort out shadowing effects caused by

the mountains and the fact that the plant has only one pyranometer and not one per

array.

A temperature correction to 25°C is also performed using temperature coefficients on power

from datasheets [9]. Finally, the efficiency is normalised with an efficiency value corresponding

to Air Mass 1.5, calculated with a fit on the overall filtered and temperature-corrected points.

In Fig. 38 the normalised efficiency is plotted against modeled average wavelength, for the

four thin-film technologies and the reference polycrystalline. Two cases are presented: data

filtered (plots on the left) and data filtered and corrected for temperature (plots on the right).

Points are grouped per different in-plane irradiance levels, around 600, 800 and 1000 W/m2. A

linear fit and the corresponding fitting equation is also displayed.

Higher irradiance levels correspond to more “blue-shifted” solar spectra, while lower irradiance

levels span a broader range. In any case, irradiance seems not to substantially affect the

slope of efficiency cloud.

56

57

Fig. 38: Normalised efficiency against average wavelength, for clear sky conditions, angle of

incidence less than 50°, sorted for shading (left) and filtered and corrected for standard

temperature of 25°C (right). Different in-plane irradiance levels are displayed.

The results show that in general the efficiency of thin-film technologies, corrected for

temperature, decreases at increasing wavelength of the solar spectrum. This effect is more

evident in amorphous-silicon, where it has to be pointed out that the Staebler-Wronski effect

plays a certain role. (It has been demonstrated that this effect has a slightly lower influence

than spectral effects [3].) Temperature correction is important especially in the case of

Cadmium Telluride, where the efficiency trend changes drastically—from increasing with

spectrum wavelength to decreasing with spectrum wavelength.

By knowing quantitatively the spectral effects on different technologies it is possible to

compare indoor with outdoor data, namely electrical parameters and temperature coefficients.

4.5.4 Acknowledgements

This work was financed through the project ”PV Initiative” in cooperation with the autonomous

Italian province of Bolzano-Alto Adige, and through the project “PV-Alps”, Interreg program IV

Italy - Swiss by the European Funds for Regional Development (EFRE).

References for this section

[1] C. Cornaro, A. Andreotti, Influence of average photon energy index on solar irradiance characteristics and

outdoor performance of photovoltaic modules, Progress in Photovoltaics: Research and Applications, (2012),

pp. 996-1003.

[2] A. Virtuani, L. Fanni, Seasonal power fluctuations of amorphous silicon thin-film solar modules: distinguishing

between different contributions, Progress in Photovoltaics: Research and Applications, (2012), pp. 208-217.

[3] R. Gottschalg, D.G. Infield, M.J. Kearney, Experimental study of variations of the solar spectrum of relevance to

thin-film solar cells, Solar Energy Materials and solar cells, Volume 79, (Sep. 2003), pp. 527-537.

[4] B. Marion, Influence of Atmospheric Variations on Photovoltaic Performance and Modeling Their Effects for

Days with Clear Skies, 38th IEEE PVSC, Austin, TX, USA, (2012), pp. 003402-003407.

[5] R. Mueller, T. Behrendt, A. Hammer, A. Kemper, A new algorithm for the satellite-based retrieval of solar

surface irradiance in spectral bands, Remote Sensing, Volume 4, (Sep. 2012), pp. 622-647.

[6] S. Kato, T. Ackerman, J. Mather, E. Clothiaux, The k-distribution method and correlated-k approximation for a

short-wave radiative transfer, Journal of Quantitative Spectroscopy and Radiative Transfer, Volume 62, (1999),

pp. 109-121.

[7] Meteoswiss: CM SAF Data and Methods [Online]. Available at

http://meteosuisse.ch/web/en/research/current_projects/climate/cmsaf/data_methods.html[Accessed: 26-Feb-2012].

58

[8] G. Belluardo, M. Pichler, D. Moser, M. M. Nikolaeva-Dimitrova, One-year comparison of different thin-film

technologies at Bolzano Airport Test Installation. Fuelling the Future: Advances in Science and Technologies for

Energy Generation, Transmission and Storage, (2012), pp. 229 – 234.

[9] B. Marion, A methodology for modeling the current-voltage curve of a PV module for outdoor conditions,

Progress in Photovoltaics: Research and Applications, Vol. 10, (2002), pp. 205-214.

59

For further information about the IEA – Photovoltaic Power Systems Programme and Task 13

publications, please visit www.iea-pvps.org.

60

61

1.7868